WO2022080367A1 - 予測方法、予測装置 - Google Patents
予測方法、予測装置 Download PDFInfo
- Publication number
- WO2022080367A1 WO2022080367A1 PCT/JP2021/037735 JP2021037735W WO2022080367A1 WO 2022080367 A1 WO2022080367 A1 WO 2022080367A1 JP 2021037735 W JP2021037735 W JP 2021037735W WO 2022080367 A1 WO2022080367 A1 WO 2022080367A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- resin powder
- prediction
- infrared
- water
- measurement data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/44—Resins; Plastics; Rubber; Leather
- G01N33/442—Resins; Plastics
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N2021/3595—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using FTIR
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2201/00—Features of devices classified in G01N21/00
- G01N2201/12—Circuits of general importance; Signal processing
- G01N2201/126—Microprocessor processing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2201/00—Features of devices classified in G01N21/00
- G01N2201/12—Circuits of general importance; Signal processing
- G01N2201/129—Using chemometrical methods
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2201/00—Features of devices classified in G01N21/00
- G01N2201/12—Circuits of general importance; Signal processing
- G01N2201/129—Using chemometrical methods
- G01N2201/1296—Using chemometrical methods using neural networks
Definitions
- the present disclosure relates to a prediction method and a prediction device for predicting the physical properties of a water-absorbent resin powder.
- the water-absorbent resin (Super Absorbent Polymer, hereinafter abbreviated as "SAP") is a resin having water swellability and water insolubility. SAP is often in the form of powder (or particles).
- SAP water absorption ratio
- AAP water absorption ratio under load
- SFC physiological saline flow inducibility
- Patent Document 1 discloses a method of predicting the physical properties of a water-absorbent resin by using a specific Raman spectrum for the water-absorbent resin.
- the target sample is irradiated with a specific single wavelength, and the scattered light with a wave number in a specific range is measured. Due to such characteristics of irradiation light, the Raman spectrum is not easily affected by the particle size of the measurement target. Therefore, the Raman spectrum is not suitable for accurately measuring the particle size to be measured.
- the target sample is irradiated with a plurality of wavelengths or a continuous spectrum of near-infrared light (generally, a wavelength of 750 nm to 2500 nm), and transmission, absorption, refraction, reflection, and diffusion are performed.
- a measurement including information on the particle size of the target sample it is possible to perform a measurement including information on the particle size of the target sample.
- the measurement can be performed in a wide wavelength range as compared with the Raman spectrum measurement, so that more accurate average information of the target sample can be obtained.
- One aspect of the present disclosure is to realize the prediction of the physical properties of the water-absorbent resin powder from the near-infrared absorption spectrum in the manufacturing process of the water-absorbent resin powder.
- the prediction method is a method for predicting the physical properties of the resin powder (Note: the resin powder is a water-absorbent resin powder and the water-absorbent resin).
- the measurement data acquisition step for acquiring the near-infrared measurement data showing the near-infrared absorption spectrum of the resin powder, and the near-red
- At least one or more of the outside measurement data and one or more processing data generated based on the near-infrared measurement data is input to the prediction model, and the prediction information related to the physical properties of the resin powder is output.
- the prediction information related to the physical properties of the resin powder is output.
- the prediction device is a prediction device for predicting the physical properties of the resin powder (Note: the resin powder is a water-absorbent resin powder and the said. (Points to any of the intermediate products produced in the manufacturing process for producing the water-absorbent resin powder), measurement data acquisition unit (acquires measurement data showing the near-infrared absorption spectrum measured for the resin powder). And, at least one of the prediction unit (the near-infrared measurement data and one or more processing data generated based on the near-infrared measurement data) is input to the prediction model to obtain the physical properties of the resin powder. Outputs related forecast information), and.
- the physical properties of the water-absorbent resin powder can be predicted from the near-infrared absorption spectrum.
- FIG. 1 is a block diagram showing an example of the configuration of the prediction system 1000.
- the prediction system 1000 includes a prediction device 100, a near-infrared spectrophotometer 3, and an external device 4.
- the prediction device 100 includes a CPU 1 and a memory 2. As shown in FIG. 1, the prediction device 100 may be communicably connected to the near-infrared spectrophotometer 3 and the external device 4.
- the communication between the prediction device 100 and the near-infrared spectrophotometer 3 may be any of short-range wireless communication, wired connection, and communication via a network such as the Internet. Alternatively, the communication between the prediction device 100 and the near-infrared spectrophotometer 3 may be configured to be directly connected by a connector such as a USB terminal.
- the communication between the prediction device 100 and the external device 4 is the same as the communication between the prediction device 100 and the near-infrared spectrophotometer 3.
- FIG. 1 shows a case where each of the near-infrared spectrophotometer 3 and the external device 4 communicably connected to the prediction device 100 is one, but the present invention is not limited to this.
- the near-infrared spectrophotometer 3 and the external device 4 which are communicably connected to the prediction device 100 may be one or a plurality of each.
- the prediction device 100 is at least one of a near-infrared measurement data showing a near-infrared absorption spectrum acquired from the near-infrared spectrophotometer 3 and one or more processed data generated based on the near-infrared measurement data. Is input to the prediction model, and prediction information related to the physical properties of the resin powder is output.
- “measurement data indicating a near-infrared absorption spectrum” may be simply referred to as a near-infrared absorption spectrum (near-infrared measurement data).
- the near-infrared spectrophotometer 3 measures the reflected light from the resin powder and the transmitted light of the resin powder when the resin powder is irradiated with near-infrared rays, and represents the near-infrared absorption characteristics of the resin powder. It is a device that calculates the external absorption spectrum.
- the near infrared ray is light having a wavelength range of 750 to 2500 nm. The near-infrared absorption spectrum will be described later.
- the external device 4 may be any device that receives the prediction result output from the prediction device 100.
- the external device 4 may be an arbitrary display device or a computer used by a manager who manages a resin powder manufacturing process.
- the external device 4 may be any manufacturing device that executes the processing of the resin powder manufacturing process.
- FIG. 2 is a functional block diagram showing an example of the main configuration of the prediction device 100.
- the prediction model 22 may be generated by a machine learning process using at least one of the following (1) and (2) as training data.
- the trained prediction model 22 may be introduced in the prediction device 100 in advance.
- the prediction device 100 may further include a function of performing machine learning processing using at least one of the above (1) and (2) as training data.
- the prediction device 100 can accurately predict the physical properties of the resin powder whose near-infrared absorption spectrum has been measured from the near-infrared absorption spectrum. can.
- the method of generating the prediction model 22 will be described later.
- the prediction device 100 outputs prediction results to a control unit 10 that collectively controls each part of the prediction device 100, a storage unit 20 that stores various data used by the control unit 10, and an external device 4.
- a communication unit 50 for outputting is provided.
- the control unit 10 corresponds to the CPU 1 of FIG. 1
- the storage unit 20 corresponds to the memory 2 of FIG.
- the communication unit 50 is for performing data communication with the external device 4.
- the communication between the prediction device 100 and the external device 4 may be any of short-range wireless communication, wired connection, and communication via a network such as the Internet.
- the prediction device 100 and the external device 4 may be directly connected by a connector such as a USB terminal.
- the control unit 10 includes a measurement data acquisition unit 11 and a prediction unit 13.
- the measurement data acquisition unit 11 acquires a near-infrared absorption spectrum from the near-infrared spectrophotometer 3.
- the measurement data acquisition unit 11 may store the acquired near-infrared absorption spectrum in the storage unit 20 as near-infrared measurement data (not shown). Further, the measurement data acquisition unit 11 may read the near-infrared absorption spectrum stored in the past from the near-infrared measurement data and use it for the subsequent prediction.
- the prediction unit 13 inputs the near-infrared absorption spectrum into the prediction model 22 described later, and outputs prediction information related to the physical properties of the resin powder.
- the prediction unit 13 may generate one or more processed data based on the near-infrared absorption spectrum.
- the processed data is data obtained by subjecting the near-infrared absorption spectrum to one or more predetermined pretreatments, unlike the raw data of the near-infrared absorption spectrum.
- the prediction unit 13 may perform preprocessing specified in the prediction model 22 on the near-infrared absorption spectrum acquired by the measurement data acquisition unit 11 (preprocessing step).
- the pretreatment includes at least one of the following processes.
- the outlier removal treatment is a near-infrared absorption that is significantly different from other near-infrared absorption spectra when the near-infrared absorption spectra measured at multiple points of the resin powder are compared with each other. This is a process for detecting a spectrum and removing the near-infrared absorption spectrum. "Measured at a plurality of places of the resin powder" corresponds to irradiating a plurality of different regions of a sample to be measured having a predetermined area made of the resin powder with measurement light.
- Specific outlier detection methods include One Class support vector machine processing (One Class SVM processing), detection using Mahalanobis distance, LOF (Local Outlier factor method), Tukey method, and neighbor method. Be done.
- -Averaging process The averaging process is a process of calculating one average spectrum data from a plurality of near-infrared absorption spectra measured at a plurality of locations of the resin powder.
- ⁇ Wavelength selection processing The wavelength selection process is a process of selecting the wavelength range of the spectral data to be input to the prediction model 22 described later. In the wavelength range processing, for example, a wavelength range in which a characteristic absorption pattern appears for each resin powder whose near-infrared absorption spectrum has been measured may be selected.
- the differential processing generates differential data obtained by differentiating spectral data with respect to wavelength.
- the differential data may include data obtained by first-order differentiating the spectral data with respect to wavelength and data obtained by second-order differentiation.
- -Baseline correction processing is a processing for aligning the baselines of a plurality of near-infrared absorption spectra measured at a plurality of locations of the resin powder.
- the prediction unit 13 may perform the following processing on the near-infrared absorption spectrum.
- ⁇ Smoothing processing weighted moving average processing, smoothing spline processing, etc.
- SNV Standard Normal Variate
- MSC Multiple scattering correction processing
- PCA Principal component analysis
- near-infrared measurement data showing the near-infrared absorption spectra of the resin powder is acquired at a plurality of locations of the resin powder, and the acquired plurality of near-infrared absorption spectra are obtained. It may include an averaging step of performing an averaging process and calculating average spectral data. In the prediction step, the average spectral data may be input to the prediction model 22 as processing data. Further, the averaging step may be a process defined by the prediction model 22.
- the prediction method performed by the prediction unit 13 may further include, for example, a wavelength range selection step of selecting the wavelength range of the average spectrum data input to the prediction model 22. Further, in the prediction step, the average spectral data in the wavelength range may be input to the prediction model 22 as processing data. Further, the wavelength range selection step may be a process defined by the prediction model 22.
- the prediction method performed by the prediction unit 13 may further include, for example, a differential data generation step of generating differential data obtained by differentiating the average spectral data in the wavelength range described above with respect to the wavelength. Further, in the prediction step, the differential data may be input to the prediction model 22 as processing data. Further, the differential data generation step may be a process defined by the prediction model 22.
- FIG. 3 is a flowchart showing the flow of processing performed by the prediction device 100.
- the measurement data acquisition unit 11 acquires near-infrared measurement data, which is a near-infrared absorption spectrum measured by the near-infrared spectrophotometer 3 (step S1: near-infrared measurement data acquisition step).
- the prediction unit 13 reads the prediction model 22 from the storage unit 20 (step S2).
- the prediction unit 13 inputs the near-infrared measurement data acquired in step S1 into the prediction model 22 (step S3). At this time, the prediction unit 13 may perform the above-mentioned pretreatment on the acquired near-infrared absorption spectrum based on the prediction model 22. The prediction unit 13 may perform one of the above-mentioned pretreatments, or may perform two or more of them. The preprocessing performed by the prediction unit 13 will be described later with a specific example.
- the prediction unit 13 predicts the physical properties of the prediction target from the pre-processed near-infrared measurement data or the unprocessed near-infrared measurement data based on the prediction model 22 (step S4: prediction step).
- the communication unit 50 outputs the prediction information indicating the prediction result output from the prediction unit 13 to the external device 4 (step S4).
- Example of pretreatment> a specific pretreatment performed by the prediction unit 13 when predicting the gel D50 from the near-infrared absorption spectrum of the hydrogel, which is an intermediate product in the manufacturing process for manufacturing the resin powder, is taken as an example. I will explain it by citing it.
- the prediction unit 13 performs outlier detection processing (for example, One Class SVM, etc.) on a plurality of near-infrared absorption spectra acquired by the measurement data acquisition unit 11, and is significantly deviated from other near-infrared absorption spectra. Remove the near-infrared absorption spectrum.
- outlier detection processing for example, One Class SVM, etc.
- the prediction unit 13 performs an averaging process on the remaining plurality of near-infrared absorption spectra to generate one average spectrum data.
- the pretreatment performed by the prediction unit 13 differs depending on which stage of the manufacturing process for producing the resin powder the measurement target of the near-infrared absorption spectrum is, and what the physical properties of the prediction target are. May be good. That is, the prediction unit 13 may perform wavelength selection processing for selecting the wavelength range of the spectrum data. Alternatively, the prediction unit 13 may perform a differential process for generating differential data obtained by differentiating the spectral data with respect to the wavelength. Further, these processes may be performed in combination.
- the prediction accuracy of the prediction device 100 can be improved.
- FIG. 4 is a functional block diagram showing an example of the main part configuration of the prediction device 100 that generates the prediction model 22.
- the prediction device 100 may generate the prediction model 22 by performing arbitrary known supervised machine learning.
- the control unit 10 includes a measurement data acquisition unit 11, a prediction unit 13, and a prediction model generation unit 18.
- the measurement data acquisition unit 11 acquires a plurality of near-infrared absorption spectra (also referred to as near-infrared absorption spectrum groups) included in the near-infrared measurement data 21 designated by the prediction model generation unit 18, and the near-infrared absorption spectrum.
- the absorption spectrum group is output to the prediction unit 13.
- the prediction unit 13 reads the prediction model candidate (described later) generated by the prediction model generation unit 18 from the prediction model generation unit 18. Further, the near-infrared absorption spectrum group included in the near-infrared measurement data 21 designated by the prediction model generation unit 18 is input to the prediction model candidate, and the physical property group corresponding to the input near-infrared absorption spectrum group is predicted. The predicted result is output to the prediction model generation unit 18.
- the predictive model generation unit 18 generates predictive model candidates for learning / verification by machine learning.
- a predictive model candidate is a predictive model for which prior machine learning has not been completed.
- the prediction model candidate is stored in the storage unit 20 as the prediction model 22. Further, the prediction model generation unit 18 designates a data group to be used for machine learning from the near-infrared measurement data 21 and the physical property information 23 possessed by the storage unit 20.
- the predictive model generation unit 18 may calculate a model evaluation index by comparing (1) and (2) below.
- the model evaluation index is, for example, an index for evaluating an error between the prediction result of (1) and the physical property group included in the physical property information 23 of (2).
- the model evaluation index may be any index capable of evaluating the accuracy of the prediction result, may be a mean square error, or may be a coefficient of determination (R 2 ).
- the predictive model generation unit 18 determines whether or not the predictive model candidate satisfies a predetermined evaluation standard based on the model evaluation index.
- the predetermined evaluation standard is a standard arbitrarily determined in advance for evaluating the prediction accuracy of the prediction model candidate.
- the predictive model generation unit 18 stores the predictive model candidate in the predictive model 22 as an optimum predictive model. On the other hand, when the generated predictive model candidate does not meet the predetermined evaluation criteria, the predictive model generation unit 18 updates the predictive model candidate.
- Update of predictive model candidate is to update the predictive model candidate by updating the weight, hyperparameters, etc. of the predictive model candidate so that the error between the prediction result and the physical property included in the physical property information 23 is minimized. And may include generating new predictive model candidates. An error backpropagation method or the like may be applied to update the predictive model candidate.
- the physical property information 23 includes the physical property information of the final product associated with the near-infrared measurement data including the near-infrared absorption spectra of a plurality of manufactured resin powders having known physical properties that have been manufactured in the past. Further, the physical property information 23 is provided in the near-infrared measurement data 21 showing the near-infrared absorption spectra of a plurality of produced intermediate products having known physical properties, which are generated in the manufacturing process for producing each manufactured resin powder. Contains physical property information of the associated intermediate product. The physical property information may be information related to the physical properties of the water-absorbent resin powder, which will be described later.
- the physical property information 23 may include, as the physical properties, the actually measured value of the resin powder to be measured or the actually measured value of the intermediate product. Further, a measurement ID may be assigned to each physical property.
- the physical property information 23 may include a physical property group used for machine learning to generate a prediction model 22 from a prediction model candidate.
- the near-infrared measurement data 21 includes a data file of the near-infrared absorption spectrum of the resin powder or the intermediate product to be measured.
- This data file can be, for example, a csv file and a text file.
- a measurement ID may be assigned to each data file of the near-infrared absorption spectrum.
- the near-infrared measurement data 21 may include a group of near-infrared absorption spectra used for machine learning to generate a prediction model 22 from a prediction model candidate.
- FIG. 5 shows the data structure of the near-infrared measurement data 21, and FIG. 6 shows the data structure of the physical property information 23.
- the near-infrared measurement data 21 has a plurality of near-infrared absorption spectrum data files, and a measurement ID is assigned to each near-infrared absorption spectrum data file.
- the physical property information 23 has a plurality of data files of physical properties (measured values), and a measurement ID is assigned to each physical property.
- this measurement ID may be given the same ID as the measurement ID given to the data file of the near-infrared absorption spectrum described above, and the near-infrared absorption spectra of the same IDs and the near-infrared absorption spectra thereof.
- the physical property information may be the result of measuring the same product. For example, the near-infrared absorption spectrum to which the measurement ID "001" in FIG. 5 is assigned and the physical characteristics to which the measurement ID "001" in FIG. 6 is assigned were measured for the same resin powder or intermediate product. It may be data.
- the prediction unit 13 uses the prediction model candidate specified by the prediction model generation unit 18.
- the prediction unit 13 acquires the measurement ID of the near-infrared absorption spectrum designated by the prediction model generation unit 18, and corresponds to the near-infrared absorption spectrum group read from the storage unit 20 having the same measurement ID as the measurement ID. You may compare the physical characteristics of the product with the information about the same product. Alternatively, the prediction result output from the prediction unit 13 may be given the same measurement ID as the measurement ID of the near-infrared absorption spectrum input to the prediction model candidate.
- the prediction model generation unit 18 generates the same prediction result output from the prediction unit 13 and a physical property group corresponding to the near-infrared absorption spectrum group read from the storage unit 20 having the same measurement ID as the prediction result. It may be compared as information about an object.
- the measurement ID given to the near-infrared absorption spectrum and the measurement ID given to the physical properties may be different.
- the measurement ID of the physical properties and the measurement ID of the near-infrared absorption spectrum may be associated with each other.
- the prediction model 22 may be generated by using either linear regression or non-linear regression machine learning.
- machine learning to generate the prediction model 22 for example, as linear regression, PLS (partial least squares regression), PCR (principal component regression), simple regression, multiple regression, ridge regression, lasso regression, Bayesian linear regression, etc. Can be mentioned.
- non-linear regression include (convolutional) neural networks, support vector regression, k-nearest neighbors, regression trees, and the like. Ensemble learning that combines the methods listed above may be used.
- the prediction model 22 may be a model that numerically predicts various physical property values, or may be a model that makes a judgment prediction as to whether the physical property values pass or fail.
- PLS and PCR machine learning are used to generate the prediction model 22.
- the prediction model 22 may define the necessary preprocessing to be performed on the near-infrared absorption spectrum.
- FIG. 7 is a flowchart showing a flow of processing performed by the prediction device 100 that executes machine learning.
- the prediction device 100 uses a combination of the near-infrared measurement data 21 and the physical property information 23 corresponding to the near-infrared measurement data 21 as learning data to generate the prediction model 22. It will be explained by listing in.
- the measurement data acquisition unit 11 reads from the storage unit 20 the near-infrared absorption spectrum group included in the near-infrared measurement data 21 designated by the prediction model generation unit 18 for use as a prediction model candidate. Further, the measurement data acquisition unit 11 reads the physical property group associated with the near-infrared absorption spectrum group included in the physical property information 23 (step S11).
- the prediction model generation unit 18 generates a prediction model candidate and outputs the prediction model candidate to the prediction unit 13 (step S12).
- the prediction unit 13 inputs the near-infrared absorption spectrum group acquired by the measurement data acquisition unit 11 into the prediction model candidate (step S13).
- the prediction unit 13 outputs the prediction result of predicting the physical property group corresponding to the near-infrared absorption spectrum group input to the prediction model candidate (step S14).
- the prediction model generation unit 18 calculates a model evaluation index by comparing the physical property group associated with the input near-infrared absorption spectrum group with the prediction result output from the prediction unit 13 (step). S15).
- the predictive model generation unit 18 determines whether or not the predictive model candidate satisfies a predetermined evaluation criterion based on the model evaluation index (step S16). When the predictive model candidate satisfies a predetermined evaluation criterion (YES in step S16), the predictive model generation unit 18 stores the predictive model candidate as the optimum predictive model candidate in the predictive model 22 (step S19). ..
- the predictive model generation unit 18 updates the predictive model candidate (step S12).
- the prediction model generation unit 18 may update the weights, hyperparameters, etc. of the prediction model candidates that do not satisfy the evaluation criteria, or may generate new prediction model candidates.
- steps S12 to S16 are repeated until step S16 becomes YES.
- the prediction model generation unit 18 may generate a plurality of predictive model candidates.
- the prediction model candidate having the highest prediction accuracy may be stored in the prediction model 22 as the optimum prediction model candidate.
- machine learning may be performed as a prediction model including preprocessing.
- the method for measuring the near-infrared absorption spectrum of the present disclosure is a method for measuring the near-infrared absorption spectrum of the resin powder for use in the prediction method by the prediction device 100 described above.
- the measuring method includes a step of irradiating the resin powder with near infrared rays and a step of calculating the near infrared absorption spectrum of the resin powder from the measured values obtained by measuring at least one of the reflected light and the transmitted light from the resin powder. And include.
- the resin powder is either a water-absorbent resin powder or an intermediate product produced in a manufacturing process for producing the water-absorbent resin powder.
- the near-infrared absorption spectrum will be described below.
- the near-infrared absorption spectrum is measured by irradiating a sample with near-infrared rays in a specific wavelength region and using near-infrared spectroscopy to detect transmitted light or reflected light.
- the near-infrared absorption spectrum can be measured by, for example, a near-infrared spectrophotometer.
- the near-infrared spectrophotometer is not particularly limited, but for example, FT-NIR NIRFlex (trademark registration) N-500 series and NIRMaster series (manufactured by BUCHI), IRMA51 series and IRMD51 series (manufactured by Chino Co., Ltd.), IR Tracer100.
- a NIR system manufactured by Shimadzu Corporation
- a Spectrom3 NIR manufactured by PerkinElmer
- a MATRIX series FT-NIR spectrometer manufactured by BRUKER
- Commercially available software can be used for analysis of the obtained spectral data.
- the near-infrared spectrophotometer may have different names depending on the manufacturer such as a near-infrared multi-component analyzer or a near-infrared analyzer.
- Near infrared rays are light having a wavelength belonging to the wavelength region of 750 to 2500 nm.
- the near-infrared spectrum is measured by irradiating light including near-infrared rays having a wavelength belonging to the wavelength region.
- the wavelength of the irradiation light may be all wavelengths in the near infrared wavelength region, or may be one or more selected specific wavelengths.
- the above-mentioned irradiation light is applied to the water-absorbent resin to be measured, and the transmitted, absorbed, refracted, reflected, and diffused light is measured.
- the near-infrared absorption spectrum is measured at least at any time before the polymerization step, between the polymerization step and the drying step, and after the drying step, and the prediction information output in the prediction step is the prediction information. It may be used to control any one or more manufacturing devices used in the resin powder manufacturing process.
- the advantages that the near-infrared absorption spectrum can be used are (1) rapid acquisition of analysis results, (2) non-contact and non-destructive analysis, and (3) quantitative analysis of multiple components at the same time. (4) Physical quantities (particle size, etc.) can be measured, and (5) operation is easy.
- the prediction method of the present disclosure outputs prediction information related to the physical properties of at least one of the water-absorbent resin powder and the intermediate product produced in the manufacturing process for producing the water-absorbent resin powder.
- the physical properties that can be predicted by the prediction device 100 may include at least one of the following (1) to (16).
- Water-absorbent resin in the present disclosure means a water-swellable water-insoluble crosslinked polymer, and is generally in the form of particles. “Water swelling” refers to NWSP 241.0.
- the non-pressurized absorption ratio (CRC) defined by R2 (15) means that the absorption ratio (CRC) is 5 g / g or more, and “water-insoluble” means NWSP 270.0. It means that the soluble content (Ext) defined by R2 (15) is 50% by mass or less.
- the water-absorbent resin can be appropriately designed according to its use, and is not particularly limited, but may be a hydrophilic crosslinked polymer obtained by cross-linking and polymerizing an unsaturated monomer having a carboxyl group. preferable. Further, the composition is not limited to a form in which the total amount (100% by weight) is a polymer, and a composition containing a surface crosslinked product, an additive, or the like may be used as long as the above performance is maintained.
- the "water-absorbent resin” is "poly (meth) acrylic acid (salt)", and may contain (meth) acrylic acid and / or a salt thereof as a repeating unit as a main component.
- NWSP stands for "Non-Woven Standard Procedures-Edition 2015”
- EDANA European Nonwoven Fabric Industry Association
- INDA Association Society
- INDA Association Technology
- the gel D50 is a mass average particle size converted into a solid content of a hydrogel containing an intermediate product.
- Gel D50 is measured according to the method described in WO2016 / 204302.
- the gel D50 of the present disclosure is a value corresponding to Solid D50 described in WO2016 / 204302.
- the gel D50 can be measured after the polymerization step and before the drying step.
- the water-absorbent resin is produced by aqueous solution polymerization, it is measured after the gel crushing step described later or before the drying step.
- CRC is an abbreviation for Centrifuge Retention Capacity (centrifugation holding capacity), and means a water absorption ratio under no pressure of a water-absorbent resin (sometimes referred to as "water absorption ratio").
- 0.2 g of the water-absorbent resin is placed in a bag made of a non-woven fabric, and then immersed in a large excess of 0.9 wt% sodium chloride aqueous solution for 30 minutes to freely swell the water-absorbent resin, and then centrifuged. It refers to the water absorption ratio (unit: g / g) after draining with a separator (250 G).
- AAP is an abbreviation for Absorption Against Pressure, and means the water absorption ratio under pressure of the water-absorbent resin.
- AAP swells 0.9 g of a water-absorbent resin with a large excess of 0.9 wt% sodium chloride aqueous solution under a load of 2.06 kPa (21 g / cm2, 0.3 psi) for 1 hour. It refers to the water absorption ratio (unit: g / g) after that. In some cases, the load condition is changed to 4.83 kPa (49 g / cm 2 , 0.7 psi) for measurement.
- SFC is an abbreviation for (Saline Flow Condivity / Saline flow inducibility), and the permeability of a 0.69 wt% sodium chloride aqueous solution to a water-absorbent resin at a load of 2.07 kPa (unit; ⁇ 10- ) . 7 ⁇ cm 3 ⁇ s ⁇ g -1 ).
- SFC is measured according to the SFC test method disclosed in US Pat. No. 5,669,894.
- T20 is the water absorption time, which is the time (unit: seconds) required for 1 g of the resin powder to absorb 20 g of the 0.9 wt% sodium chloride aqueous solution, and is disclosed in US Publication of Patent Publication US2012 / 0318046. It is measured according to the measured method.
- U20> "U20" is absorption in 20 minutes (unit; g / g) and is measured according to the measuring method disclosed in US Publication of Patent Publication US2012 / 0318046.
- K20 is the effective transmittance (unit: m 2 ) in 20 minutes, and is measured according to the measuring method disclosed in US Publication of Patent Publication US2012 / 0318046.
- Vortex water absorption time
- 0.02 part by mass of edible blue No. 1 brilliant blue
- a pre-adjusted physiological saline solution 0.9 mass% sodium chloride aqueous solution
- D50 is the mass average particle size of the resin powder produced in the drying step, which will be described later.
- the mass average particle size (D50) is the same as that described in US Pat. No. 7,638,570, “(3) Mass-Average Particle Diameter (D50) and Logarithmic Standard deviation Method of Particle Diameter”.
- the water content and solid content of the water-containing gel are the water content and resin solid content of the water-containing gel before drying.
- the water content and solid content of the water-containing gel can be measured after the polymerization step and before the drying step. That is, it may be the water content and solid content of the water-containing gel before being crushed, or it may be the water content and solid content of the particulate water-containing gel after crushing.
- the water content of the water content gel is measured according to NWSP.
- the mass of the sample is changed to 2.0 g
- the drying temperature is changed to 180 ° C.
- the drying time is changed to 24 hours.
- the total mass W1 (g) of the sample hydroogen gel and the aluminum cup
- the sample is allowed to stand in an oven set to an atmospheric temperature of 180 ° C. After 24 hours, the sample is removed from the oven and the total mass W2 (g) is accurately weighed.
- the mass of the hydrous gel used in this measurement is M (g)
- the water content (100- ⁇ ) (mass%) of the hydrogel is determined according to the following (formula 1).
- ⁇ is a solid content ratio (mass%) of a hydrogel.
- Gel CRC is a CRC of a hydrogel before drying.
- the gel CRC can be measured after the polymerization step and before the drying step. That is, it may be the CRC of the hydrogel before pulverization, or the CRC of the particulate hydrogel after pulverization.
- gel CRC means that 0.6 g of water-containing gel is placed in a non-woven fabric bag and then immersed in a large excess of 0.9 wt% sodium chloride aqueous solution for 24 hours to freely release the water-absorbent resin. It refers to the water absorption ratio (unit: g / g) after swelling and then draining with a centrifuge (250 G).
- Ext is an abbreviation for Extremes and means a water-soluble component (amount of water-soluble component). Specifically, the amount of the dissolved polymer (unit:% by weight) after adding 1.0 g of the water-absorbent resin to 200 mL of a 0.9 wt% sodium chloride aqueous solution and stirring for 16 hours. The amount of dissolved polymer is measured using pH titration.
- Gel Ext is the Ext of the hydrogel before drying.
- the gel Ext can be measured after the polymerization step and before the drying step. That is, it may be the Ext of the hydrogel before pulverization, or it may be the Ext of the particulate hydrogel after pulverization.
- the sample is changed to 2.0 g and measured, and calculated as the mass% of the water-soluble content per solid content.
- Residal Monomers is the amount of the monomer remaining in the water-absorbent resin, and NWSP210.0. It is measured according to R2 (19).
- FSR is the water absorption rate (unit: g / g / s) and is measured according to the measuring method disclosed in International Publication No. 2009/016055.
- FSC Free Swell Capacity
- NWSP240.0 It is measured according to R2 (15).
- Flow Rate means the flow rate of the water-absorbent resin. Flow Rate is NWSP251.0. It is measured according to R2 (15).
- Density means the bulk specific density of the water-absorbent resin. Density is NWSP251.0. Measured according to R2 (15).
- the above-mentioned physical properties may be measured at any timing in the resin powder manufacturing process.
- the method for producing the water-absorbent resin powder may include a polymerization step and a drying step, and preferably includes a gel crushing step, a post-crosslinking step, and a granulation step. Each process will be described below.
- a water-containing gel-like crosslinked polymer (hereinafter referred to as a water-containing gel-like crosslinked polymer) is polymerized by polymerizing a monomer containing acrylic acid (salt) as a main component and a monomer aqueous solution containing at least one type of polymerizable internal crosslinker. , Expressed as "hydrous gel").
- the polymerization initiator used in the present disclosure is appropriately selected depending on the polymerization form and the like, and is not particularly limited.
- a thermal decomposition type polymerization initiator a photodegradable polymerization initiator, or decomposition of these polymerization initiators.
- examples thereof include a redox-based polymerization initiator used in combination with a reducing agent that promotes.
- a reducing agent that promotes e.g., one or more of the polymerization initiators disclosed in US Pat. No. 7,265,190 will be used.
- a peroxide or an azo compound is preferably used, more preferably a peroxide, and still more preferably a persulfate.
- the polymerization reaction may be carried out by irradiating with active energy rays such as radiation, electron beam, and ultraviolet rays, or these active energy rays and the polymerization initiator may be used in combination. ..
- the polymerization form applied to the present disclosure is not particularly limited, but is preferably spray droplet polymerization, aqueous solution polymerization, reverse phase suspension polymerization, etc. from the viewpoint of water absorption characteristics of the hydrogel and ease of polymerization control.
- aqueous solution polymerization a reverse phase suspension polymerization, and more preferably an aqueous solution polymerization.
- continuous aqueous solution polymerization is particularly preferable, and either continuous belt polymerization or continuous kneader polymerization is applied.
- the gel crushing step is a step of crushing the hydrous gel obtained in the polymerization step into a gel to obtain a particulate hydrogel.
- a gel crushing step may be performed at the same time as the polymerization step as in continuous kneader polymerization.
- the gel crusher during or after polymerization used in this step is not particularly limited, and is a gel crusher equipped with a plurality of rotary stirring blades such as a batch type or continuous double-arm kneader, or a single shaft.
- a gel crusher equipped with a plurality of rotary stirring blades such as a batch type or continuous double-arm kneader, or a single shaft.
- examples thereof include an extruder, a twin-screw extruder, a meat chopper, a screw-type extruder, a double-screw type kneader equipped with a crushing means, and the like.
- a screw type extruder in which a perforated plate is installed at one end of a casing is preferable, and specific examples thereof include a screw type extruder disclosed in Japanese Patent Application Laid-Open No. 2000-63527 and WO2011 / 126079. Be done.
- the gel pulverization is performed during and / or after the polymerization step, and more preferably, it is performed on the water-containing gel-like polymer after the polymerization step.
- the monomer aqueous solution continuously changes into a hydrogel-like polymer with the lapse of the polymerization time. Therefore, the hydrogel-like polymer after the time when the polymerization temperature becomes maximum, or the hydrogel-like polymer having a monomer polymerization rate of 90 mol% or more may be gel-ground.
- the maximum value of the polymerization temperature is also referred to as the polymerization peak temperature.
- the polymerization rate of the monomer is sometimes referred to as the conversion rate.
- the polymerization rate of the monomer is calculated from the amount of polymer calculated from the pH titration of the hydrogel polymer and the amount of residual monomer.
- the hydrogel-like polymer during and / or after the polymerization step preferably the hydrogel-like polymer after the polymerization step
- the hydrogel-like polymer after the polymerization step is about several tens of centimeters before gel pulverization. It can be cut or coarsely crushed to size. By this operation, it becomes easy to fill the gel-containing gel-like polymer in the gel crushing device, and the gel crushing step can be carried out more smoothly.
- the means for cutting or coarsely crushing a means capable of cutting or coarsely crushing the hydrogel-like polymer without kneading is preferable, and examples thereof include a guillotine cutter and the like.
- the size and shape of the hydrogel-like polymer obtained by the above-mentioned cutting or coarse crushing is not particularly limited as long as it can be filled in the gel crushing apparatus.
- the drying step is a step of obtaining a granular dried product by drying the particulate hydrogel to a desired solid content.
- the drying method is not particularly limited, and is, for example, heat drying, hot air drying, vacuum drying, fluidized layer drying, infrared drying, microwave drying, drum dryer drying, co-boiling dehydration with a hydrophobic organic solvent, and high temperature. High humidity drying using steam can be mentioned.
- the post-crosslinking agent described later in the main drying step may be used in the main drying step to obtain a water-absorbent resin powder that has been post-crosslinked (also referred to as surface cross-linking) in the drying step.
- the drying device used in the drying process is not particularly limited, and one or more types such as a heat transfer conduction type dryer, a radiant heat transfer type dryer, a hot air heat transfer type dryer, and a dielectric heating type dryer are used. Is selected as appropriate.
- the drying device may be a batch type or a continuous type. Further, the drying device may be a direct heating type or an indirect heating type. Further, the drying device may be any one of a static material type, a material stirring type, a material transfer type, and a hot air transfer type.
- a heat transfer type dryer such as a ventilation band type, a ventilation circuit type, a ventilation vertical type, a parallel flow band type, a ventilation tunnel type, a ventilation stirring type, a ventilation rotation type, a rotary type with a heating tube, a fluidized bed type, and an air flow type.
- a heat transfer type dryer such as a ventilation band type, a ventilation circuit type, a ventilation vertical type, a parallel flow band type, a ventilation tunnel type, a ventilation stirring type, a ventilation rotation type, a rotary type with a heating tube, a fluidized bed type, and an air flow type.
- the drying temperature in the drying step is 80 ° C. or higher, preferably 100 ° C. or higher, more preferably 120 ° C. or higher, and particularly preferably 150 ° C. or higher.
- the drying temperature is 250 ° C. or lower, preferably 230 ° C. or lower, and more preferably 220 ° C. or lower. Any combination of the upper and lower limits of the drying temperature is preferable. If the drying temperature is less than 80 ° C., the drying time until a suitable resin solid content (moisture content) is obtained becomes long, which is not preferable. In addition, undried matter is produced, and clogging may occur during the subsequent pulverization step.
- the drying temperature refers to the temperature of the heat medium used for drying in the case of direct heating, refers to the temperature of the hot air used for drying in the case of hot air drying, and the heat transfer surface used for drying in the case of indirect heating. Refers to the temperature of.
- the drying time in the drying step refers to the time until the solid content becomes 80% by weight or more, preferably 60 minutes or less, preferably 40 minutes or less, 30 minutes or less, and 25 minutes or less in that order.
- the lower limit of the drying time is about 1 minute in consideration of the drying efficiency.
- the total drying time is preferably 120 minutes or less, more preferably 100 minutes or less, 80 minutes or less, and 60 minutes or less. If the drying time is short, undried matter is produced, which may cause clogging during the subsequent grinding step.
- the particulate hydrogel obtained in the gel crushing step is dried in the drying step described above to obtain a dry polymer.
- the resin solid content obtained from the dry weight loss of the dry polymer is preferably 80% by weight or more, more preferably 85 to 99% by weight. More preferably, it is 86 to 98% by weight.
- This step is a step of adding a cross-linking agent that reacts with a functional group (particularly a carboxyl group) of a water-absorbent resin to the water-containing gel after polymerization and its dried product to cause a cross-linking reaction. Since it is mainly cross-linked from the surface of the water-absorbent resin particles, it is also called surface cross-linking or secondary cross-linking.
- a post-crosslinking agent is added to the granular hydrogel and / or the granular dried product to react.
- This step includes a post-crosslinking agent addition step and a heat treatment step, and may have a cooling step after the heat treatment step, if necessary.
- This step is a step of adjusting the particle size of the granular dry product or the post-crosslinked granular dry product.
- a water-absorbent resin powder having a more positively controlled particle size or particle size distribution can be obtained.
- the sizing step comprises a crushing step and / or a classification step.
- the crushing step is a step of crushing loosely aggregated granular dried sol with a crusher through a drying step or a heat treatment step to adjust the particle size.
- the classification step is a step of removing coarse particles and fine particles from the granular dried product, the post-crosslinked granular dried product, or their crushed product using a classifying machine.
- An ideal sizing step is to obtain a water-absorbent resin powder whose particle size and particle size distribution are controlled only by the pulverization step. Water absorption performance, handleability, and usability when applied to sanitary materials such as diapers and sanitary products vary depending on the particle size and particle size distribution of the water-absorbent resin powder. It is preferable to obtain a water-absorbent resin powder having a diameter and a particle size distribution.
- the method for producing the water-absorbent resin powder may include a cooling step, a monomer aqueous solution preparation step, a step of adding various additives, a fine powder removal step, and a fine powder recycling step. Further, other known steps may be included.
- the predictor 100 has a near-infrared absorption spectrum of an intermediate product before, between the polymerization step and the drying step, and at least after the drying step included in the manufacturing step of the water-absorbent resin powder. May be measured and predictive information related to the physical properties of the intermediate product (or the finished resin powder) at any stage of the above steps may be output.
- the production conditions may be controlled in the production process of any one or more of the water-absorbent resin powders based on the prediction information output by the prediction device 100.
- the prediction information output by the prediction device 100 may be used.
- FIG. 8 is a block diagram showing an example of the configuration of the prediction system 1000a according to another embodiment of the present disclosure.
- the prediction system 1000a includes a prediction device 100, near-infrared spectrophotometers 3a to 3f, and external devices 4a to 4e.
- the prediction device 100 is connected to the near-infrared spectrophotometers 3a to 3f and the external devices 4a to 4e.
- the external devices 4a to 4e are, for example, control devices for performing each step (polymerization step, pulverization step, etc.).
- the gel D50 is, for example, the physical characteristics of the water-absorbent resin powder measured after the gel pulverization step.
- FIG. 8 shows a case where the external device 4b is a device for controlling the gel crushing step, and the near-infrared spectrophotometer 3c is a near-infrared spectrophotometer for measuring the near-infrared absorption spectrum after the gel crushing step.
- the external device 4b is a device for controlling the gel crushing step
- the near-infrared spectrophotometer 3c is a near-infrared spectrophotometer for measuring the near-infrared absorption spectrum after the gel crushing step.
- the near-infrared spectrophotometer 3c outputs the measured near-infrared absorption spectrum of the water-absorbent resin powder to the prediction device 100.
- the prediction device 100 performs preprocessing on the acquired near-infrared absorption spectrum based on the prediction model. Examples of the pretreatment include outlier removal processing and averaging processing.
- the prediction device 100 predicts the gel D50 from the preprocessed near-infrared absorption spectrum based on the prediction model.
- the prediction device 100 outputs the prediction result to the external device 4c.
- the external device 4c may be, for example, a device that controls the drying process.
- the external device 4c that controls the drying process controls to heat the resin powder under a condition higher than the predetermined value.
- the conditions in the process may be changed.
- the external device 4b that controls the gel crushing process may increase the gel crushing load more than the predetermined value.
- the external device 4b may change the conditions in the process, such as controlling the gel to have a higher rotation speed so that higher shear is applied to the gel.
- control is performed to reduce the amount of the cross-linking agent in the polymerization step.
- control is performed to reduce the amount of the cross-linking agent in the polymerization step.
- the AAP of the final product physical properties is lower than the product standard, it is possible to control the change of the treatment agent composition in the post-crosslinking step.
- the prediction device 100 may be able to identify which near-infrared spectrophotometer has acquired the near-infrared measurement data from among the plurality of near-infrared spectrophotometers.
- the prediction device 100 acquires in advance the MAC address of each near-infrared spectrophotometer in the prediction system 1000a and the installation location of each near-infrared spectrophotometer.
- FIG. 9 shows a correspondence table between the MAC address and the near-infrared spectrophotometer.
- the prediction device can determine in which process in the prediction system 1000a the near-infrared measurement data is obtained. 100 can be identified.
- the prediction system 1000a having such a configuration is a physical property of an intermediate product in each step of producing a resin powder (for example, a crushed gel physical property such as a gel particle size) or a physical property of a resin powder as a final product (for example, described above).
- CRC and AAP can be predicted accurately in a short time. If the prediction information is used to control the manufacturing equipment (external devices 4a to 4e) that execute each process in the manufacturing process for manufacturing the resin powder, the physical property manipulation factors in each manufacturing process can be adjusted in real time, and the specifications of the out-of-spec product can be adjusted. The outbreak can be effectively suppressed.
- the near-infrared spectrophotometers 3a to 3f are generally inexpensive (at least cheaper than the Raman spectrometer). Therefore, it is possible to keep the cost for arranging a plurality of near-infrared spectrophotometers 3a to 3f in the process of manufacturing the resin powder low.
- the control block (particularly the control unit 10) of the control block (particularly the control unit 10) of the prediction device 100 according to the first embodiment and the second embodiment is a logic circuit (hardware) formed in an integrated circuit (IC chip) or the like. It may be realized by software, or it may be realized by software.
- the prediction device 100 includes a computer that executes instructions of a program that is software that realizes each function.
- the computer includes, for example, one or more processors and a computer-readable recording medium that stores the program. Then, in the computer, the processor reads the program from the recording medium and executes the program, thereby achieving the object of the present disclosure.
- the processor for example, a CPU (Central Processing Unit) can be used.
- the recording medium in addition to a “non-temporary tangible medium” such as a ROM (Read Only Memory), a tape, a disk, a card, a semiconductor memory, a programmable logic circuit, or the like can be used.
- a RAM RandomAccessMemory
- the program may be supplied to the computer via any transmission medium (communication network, broadcast wave, etc.) capable of transmitting the program. It should be noted that one aspect of the present disclosure can also be realized in the form of a data signal embedded in a carrier wave, in which the above program is embodied by electronic transmission.
- the size of the stainless steel petri dish was 88 mm in inner diameter and 20 mm in height.
- the surface temperature of the stainless steel petri dish was preheated to 50 ° C. using a hot plate (NEO HOTPLATE H1-1000, manufactured by Inuchi Seieidou Co., Ltd.).
- the stainless petri dish was covered with a glass container having an exhaust port, and the petri dish was sucked with a vacuum pump so that the pressure inside the case became 85 kPa.
- the pressure outside the case was 101.3 kPa (normal pressure).
- reaction solution (1) was poured into the stainless steel petri dish, polymerization was started shortly afterwards.
- the polymerization proceeded while expanding and foaming upward in all directions while generating water vapor, and then contracted to a size slightly larger than the bottom surface. This expansion and contraction was completed within about 1 minute.
- a hydrogel-like crosslinked polymer hereinafter referred to as “hydrogen gel”.
- the obtained hydrogel (1) was pulverized with a screw extruder (meat chopper) having the following specifications.
- the screw extruder was provided with a perforated plate at its tip, and the perforated plate had a diameter of 82 mm, a hole diameter of 8.0 mm, a number of holes of 33, and a thickness of 9.5 mm.
- the amount of the hydrous gel (1) added was about 360 g / min, and gel pulverization was performed while adding deionized water at 90 ° C. at 50 g / min in parallel with the gel addition.
- This gel-crushed particulate hydrogel (1) was used for the evaluation of gel D50, which will be described later.
- a surface cross-linking agent solution consisting of 0.025 g of ethylene glycol diglycidyl ether, 0.3 g of ethylene carbonate, 0.5 g of propylene glycol and 2.0 g of deionized water was sprayed and mixed with 100 g of the water-absorbent resin powder (1). .. The mixture was heat-treated at 200 ° C. for 35 minutes to obtain a surface-crosslinked water-absorbent resin powder (2).
- Equipment FT-NIR NIRFlex (trademark registration) N-500 (manufactured by BUCHI) Measurement wavelength: 800-2500 nm Measurement method: Diffuse reflection measurement (ii) Equipment: IRMA5184S (manufactured by Chino Co., Ltd.) Measurement wavelength (8 wavelengths): 1320, 1460, 1600, 1720, 1800, 1960, 2100, 2310 nm Measurement method: Near infrared absorption type.
- the data set used for the evaluation includes a plurality of combinations of the near-infrared measurement data and the physical properties associated with the near-infrared measurement data.
- the data set is divided into training data and verification data and used for evaluation.
- the learning data is data including a near-infrared absorption spectrum and an actually measured value of physical properties associated with the near-infrared absorption spectrum, and is data to be used for prior machine learning.
- the verification data is data that is not included in the learning data.
- the data set used for evaluation was randomly divided, and a prediction model by PLS or PCR was created for the training data.
- the near-infrared absorption spectra and physical properties of the plurality of water-absorbent resin powders (1) and (2) are measured, and the near-infrared absorption spectra and the physical properties associated with the near-infrared absorption spectra are obtained.
- FIG. 10 is a graph in which predicted values with respect to the measured values of gel D50 are plotted in the range of 80 to 190 ⁇ m.
- FIG. 11 is a graph in which the predicted values with respect to the measured values of CRC are plotted in the range of 24 to 31 g / g.
- FIG. 12 is a graph in which the predicted values with respect to the measured values of AAP are plotted in the range of 24.5 to 27 g / g.
- FIG. 13 is a graph in which predicted values with respect to the measured values of SFC are plotted in the range of 20 to 110 ( ⁇ 10 -7 ⁇ cm 3 ⁇ s ⁇ g -1 ).
- D50 The data set has 90 combinations of the near-infrared measurement absorption spectrum and the physical characteristics associated with the near-infrared absorption spectrum.
- the dataset was randomly divided into training data and verification data, 80% for training and 20% for verification.
- a prediction model by PLS was created for the training data.
- FIG. 14 is a graph in which predicted values with respect to the measured values of D50 are plotted in the range of 250 to 450 ⁇ m.
- FIG. 15 is a graph in which the predicted value of the water content with respect to the measured value is plotted in the range of 96.5 to 98.5 wt%. Since the solid content is determined by 100-moisture content (% by weight), it can be said that this graph shows a predicted value with respect to the measured value in the solid content ratio.
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Engineering & Computer Science (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Analytical Chemistry (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Food Science & Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Medicinal Chemistry (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
Description
以下、本開示の実施形態について詳細に説明する。
まず、本開示の一実施形態に係る予測装置100を備える予測システム1000の構成について図1を用いて説明する。図1は、予測システム1000の構成の一例を示すブロック図である。
以下では、予測モデル22を用いて、吸水性樹脂粉末(以下、単に「樹脂粉末」と記す場合がある)の物性を予測する予測装置100の構成について、図2を用いて説明する。図2は、予測装置100の要部構成の一例を示す機能ブロック図である。
・外れ値除去処理
外れ値除去処理は、樹脂粉末の複数個所にて測定された近赤外吸収スペクトルを相互に比較したときに、他の近赤外線吸収スペクトルから有意に外れている近赤外吸収スペクトルを検出して、該近赤外吸収スペクトルを除去するための処理である。「樹脂粉末の複数個所にて測定された」とは、樹脂粉末からなる所定の面積を有する測定対象試料の複数の異なる領域に測定光を照射して測定したことに相当する。具体的な外れ値検出方法として、One Class サポートベクトルマシン処理(One Class SVM処理)、マハラノビス距離を用いた検出、LOF(Local Outlier factor、局所外れ値因子法)、Tukey法、近傍法等が挙げられる。
・平均化処理
平均化処理は、樹脂粉末の複数個所にて測定された複数の近赤外吸収スペクトルから、1つの平均スペクトルデータを算出する処理である。
・波長選択処理
波長選択処理は、後述する予測モデル22に入力するスペクトルデータの波長範囲を選択する処理である。波長範囲処理では、例えば、近赤外吸収スペクトルが測定された樹脂粉末毎に特徴的な吸収パターンが表れている波長範囲を選択してもよい。
・微分処理
微分処理は、スペクトルデータを波長について微分した微分データを生成する。微分データは、スペクトルデータを波長について一次微分したデータ、及び二次微分したデータを含んでいてもよい。
・ベースライン補正処理
ベースライン補正処理は、樹脂粉末の複数個所にて測定された複数の近赤外吸収スペクトルのベースラインを揃える処理である。
・平滑化処理(加重移動平均処理、平滑化スプライン処理等)
・差スペクトル処理
・標準化処理(SNV(Standard Normal Variate)処理)
・多重散乱補正処理(MSC(Multiplicative Scatter Correction)処理)
・主成分分析(PCA)による次元削減
その他、分類及びクラスタリング等を行ってもよい。
以下、予測装置100が行う処理について、図3を用いて説明する。図3は、予測装置100が行う処理の流れを示すフローチャートである。
ここでは、一例として、樹脂粉末を製造する製造工程における中間生成物である含水ゲルの近赤外吸収スペクトルから、ゲルD50を予測する場合に、予測部13が行う具体的な前処理を例に挙げて説明する。
次に、予測モデル22を生成するための機械学習を行う予測装置100の構成について、図4を用いて説明する。図4は、予測モデル22を生成する予測装置100の要部構成の一例を示す機能ブロック図である。なお、説明の便宜上、図1にて説明した部材と同じ機能を有する部材については、同じ符号を付記し、その説明を繰り返さない。なお、予測装置100は、任意の公知の教師有り機械学習を行うことにより、予測モデル22を生成してもよい。
(2)記憶部20から読み込んだ物性情報23に含まれる、予測モデル候補に入力された近赤外吸収スペクトル群に対応付けられている物性群。
ここで、モデル評価指標は、例えば、(1)の予測結果と、(2)の物性情報23に含まれる物性群との誤差を評価するための指標である。モデル評価指標は、予測結果の精度を評価可能な任意の指標であってもよく、平均二乗誤差であってもよいし、決定係数(R2)であってもよい。
続いて、予測装置100が行う処理について図7を用いて説明する。図7は、機械学習を実行する予測装置100が行う処理の流れを示すフローチャートである。なお、ここでは、予測装置100が、学習データとして、近赤外測定データ21と該近赤外測定データ21に対応する物性情報23との組み合わせを使用して予測モデル22を生成する場合を例に挙げて説明する。
本開示の近赤外吸収スペクトルの測定方法は、上述した予測装置100による予測方法に用いるための樹脂粉末の近赤外吸収スペクトルを測定する方法である。前記測定方法は、樹脂粉末に近赤外線を照射するステップと、該樹脂粉末からの反射光及び透過光のうちの少なくとも一方を測定した測定値から前記樹脂粉末の近赤外吸収スペクトルを算出するステップとを含む。前記樹脂粉末は、吸水性樹脂粉末、及び該吸水性樹脂粉末を製造するための製造工程において生成される中間生成物のいずれかである。以下に、近赤外吸収スペクトルについて説明する。
ここでは、予測装置100が、樹脂粉末の物性予測に用いる近赤外吸収スペクトルについて説明する。
近赤外吸収スペクトルは、特定の波長領域の近赤外線を試料に照射して、透過光又は反射光を検出する近赤外分光法を利用して測定される。近赤外吸収スペクトルは、例えば、近赤外分光光度計により測定することができる。近赤外分光光度計としては、特に限定されないが、例えば、FT-NIR NIRFlex(商標登録)N-500シリーズ及びNIRMasterシリーズ(BUCHI製)、IRMA51シリーズ及びIRMD51シリーズ(株式会社チノー製)、IR Tracer100 NIRシステム(島津製作所製)、Spectrum3 NIR(PerkinElmer製)並びにMATRIXシリーズ FT-NIRスペクトロメータ(BRUKER製)等を使用することができる。得られたスペクトルデータの解析は、市販のソフトウェアを用いることができる。なお、近赤外分光光度計は、近赤外多成分分析計や近赤外分析計等メーカーにより名称が異なって表記されている場合がある。
近赤外線は、750~2500nmの波長領域に属する波長を有する光である。近赤外線スペクトルは、前記波長領域に属する波長を有する近赤外線を含む光を照射して測定される。照射光の波長は近赤外波長領域の全波長でもよく、選択した1つ以上の特定波長であっても良い。本開示の近赤外吸収スペクトル測定においては、上記の照射光を測定対象の吸水性樹脂に照射し、透過、吸収、屈折、反射、拡散された光を測定するため、化学的な情報だけでなく物理的な情報を収集することができる。言い換えると、測定対象試料の温度、測定光路中の雰囲気(蒸気有無、窒素置換有無、気圧など)、表面粗さ、試料厚み、試料充填状態、測定までの時間などの影響を受ける。従い、予測精度の観点からは、近赤外吸収スペクトルを取得する際は、できる限り上記の物理的条件が均一な条件下で測定することが好ましい。また必要に応じて、上記の物理的条件(例えば前記試料温度)を別途測定しておき、当該測定値に基づいて、対応する近赤外吸収スペクトルを補正してもよい。
本開示の予測方法は、吸水性樹脂粉末、及び該吸水性樹脂粉末を製造するための製造工程において生成される中間生成物の少なくともいずれかの物性に関連する予測情報を出力する。
(2)CRC
(3)AAP
(4)SFC
(5)T20、U20、K20
(6)Vortex
(7)D50
(8)含水ゲルの含水率
(9)固形分率
(10)Residual Monomers
(11)FSR
(12)FSC
(13)Flow Rate
(14)Density
(15)Ext
(16)ゲルExt
予測装置100が予測対象とし得る物性としては、(1)ゲルD50、(2)CRC、(3)AAP、(4)SFC、(6)Vortex、(7)D50、(8)含水ゲルの含水率、(9)固形分率のうち少なくとも1つであることが好ましい。
本開示おける「吸水性樹脂」とは、水膨潤性水不溶性の架橋重合体を意味し、一般的に粒子状である。また、「水膨潤性」とは、NWSP 241.0.R2(15)で規定される無加圧下吸収倍率(CRC)が5g/g以上であることを意味し、「水不溶性」とは、NWSP 270.0.R2(15)で規定される可溶分(Ext)が50質量%以下であることを意味する。
ゲルD50は、中間生成物である含水ゲルの固形分に換算した質量平均粒子径である。ゲルD50は、WO2016/204302に記載の方法に準拠して測定する。本開示のゲルD50は、WO2016/204302に記載のSolidD50に対応する値である。
「CRC」は、Centrifuge Retention Capacity(遠心分離保持容量)の略称であり、吸水性樹脂の無加圧下吸水倍率(「吸水倍率」と称する場合もある)を意味する。
「AAP」は、Absorption Against Pressureの略称であり、吸水性樹脂の加圧下吸水倍率を意味する。
「SFC」は、(Saline Flow Conductivity/生理食塩水流れ誘導性)の略称であり、荷重2.07kPaでの吸水性樹脂に対する0.69重量%塩化ナトリウム水溶液の通液性(単位;×10-7・cm3・s・g-1)をいう。「SFC」は、米国特許第5669894号に開示されたSFC試験方法に準じて測定される。
「T20」は、吸水時間のことであり、1gの樹脂粉末が0.9重量%塩化ナトリウム水溶液20gを吸収するために要する時間(単位;秒)であり、米国公開特許公報US2012/0318046に開示された測定方法に準拠して測定される。
「U20」は、20分での吸収(単位;g/g)であり、米国公開特許公報US2012/0318046に開示された測定方法に準拠して測定される。
「K20」は、20分での有効透過率(単位;m2)であり、米国公開特許公報US2012/0318046に開示された測定方法に準拠して測定される。
Vortex(吸水時間)は、以下の手順にしたがって測定する。先ず、予め調整された生理食塩水(0.9質量%塩化ナトリウム水溶液)1000質量部に、食品添加物である食用青色1号(ブリリアントブルー)0.02質量部を添加した後、液温を30℃に調整する。
本開示において、「D50」は、後述する、乾燥工程において製造された樹脂粉末の質量平均粒子径である。質量平均粒子径(D50)は、米国特許第7638570号に記載された「(3)Mass-Average Particle Diameter(D50)and Logarithmic Standard Deviation of Particle Diameter Distribution」と同様の方法で測定する。
含水ゲルの含水率及び固形分率は、乾燥前の含水ゲルの含水率及び樹脂固形分率のことである。含水ゲルの含水率及び固形分率は、重合工程の後から乾燥工程の前までにおいて測定することができる。すなわち、粉砕される前の含水ゲルの含水率及び固形分率であってもよく、粉砕後の粒子状含水ゲルの含水率及び固形分率であってもよい。
「ゲルCRC」は、乾燥前の含水ゲルのCRCである。ゲルCRCは、重合工程の後から乾燥工程の前までにおいて測定することができる。すなわち、粉砕される前の含水ゲルのCRCであってもよく、粉砕後の粒子状含水ゲルのCRCであってもよい。
「Ext」は、Extractablesの略称であり、水可溶分(水可溶成分量)を意味する。具体的には、吸水性樹脂1.0gを0.9重量%塩化ナトリウム水溶液200mLに添加し、16時間撹拌した後の溶解ポリマー量(単位;重量%)である。溶解ポリマー量の測定は、pH滴定を用いて行う。
「ゲルExt」は、乾燥前の含水ゲルのExtである。ゲルExtは、重合工程の後から乾燥工程の前までにおいて測定することができる。すなわち、粉砕される前の含水ゲルのExtであってもよく、粉砕後の粒子状含水ゲルのExtであってもよい。
「Residual Monomers」は、吸水性樹脂中に残存する単量体(モノマー)量であり、NWSP210.0.R2(19)に準拠して測定される。
「FSR」は、吸水速度であり(単位;g/g/s)、国際公開第2009/016055号に開示された測定方法に準拠して測定される。
「FSC」は、Free Swell Capacityの略称であり、吸水性樹脂の無加圧下吊り下げ吸水倍率を意味する。FSCは、NWSP240.0.R2(15)に準拠して測定される。
「Flow Rate」は、吸水性樹脂の流下速度を意味する。Frow Rateは、NWSP251.0.R2(15)に準拠して測定される。
「Density」は、吸水性樹脂の嵩比重を意味する。Densityは、NWSP251.0.R2(15)に準拠し測定される。
上述した、樹脂粉末の各物性は、吸収性樹脂粉末の製造工程内において測定されるものである。以下に、樹脂粉末の製造方法について説明する。
重合工程は、一例として、アクリル酸(塩)を主成分として含む単量体、及び少なくとも1種類の重合性内部架橋剤を含む単量体水溶液を重合させて、含水ゲル状架橋重合体(以下、「含水ゲル」と表記する)を得る工程である。
本開示で使用される重合開始剤は、重合形態等によって適宜選択されるため、特に限定されないが、例えば、熱分解型重合開始剤、光分解型重合開始剤、又はこれらの重合開始剤の分解を促進する還元剤を併用したレドックス系重合開始剤等が挙げられる。具体的には、米国特許第7265190号に開示された重合開始剤のうちの、一種類又は二種類以上が用いられる。尚、重合開始剤の取り扱い性や粒子状吸水剤又は吸水性樹脂の物性の観点から、好ましくは過酸化物又はアゾ化合物、より好ましくは過酸化物、さらに好ましくは過硫酸塩が使用される。
本開示に適用される重合形態としては、特に限定されないが、含水ゲルの吸水特性や重合制御の容易性等の観点から、好ましくは噴霧液滴重合、水溶液重合、逆相懸濁重合、より好ましくは水溶液重合、逆相懸濁重合、さらに好ましくは水溶液重合が挙げられる。中でも、連続水溶液重合が特に好ましく、連続ベルト重合、連続ニーダー重合の何れでも適用される。
ゲル粉砕工程は、重合工程で得られた含水ゲルをゲル粉砕して、粒子状含水ゲルを得る工程である。噴霧液滴重合、逆相懸濁重合により吸水性樹脂を製造する場合は粒子状含水ゲルを得ることができるので、ゲル破砕工程を行わなくてもよい。また連続ニーダー重合のように重合工程と同時にゲル破砕工程を行ってもよい。特に、吸水速度の高いSAPを得る観点からは、本ゲル粉砕工程において、含水ゲルを細粒化し、ゲルD50が所望の範囲である粒子状含水ゲルを製造することが好ましい。
本工程で使用される重合時又は重合後のゲル粉砕装置としては、特に限定されず、バッチ型又は連続型の双腕型ニーダー等、複数の回転攪拌翼を備えたゲル粉砕機や、1軸押出機、2軸押出機、ミートチョッパー、スクリュー型押出機、破砕手段を備えた複軸型混錬機等が挙げられる。
本開示において上記ゲル粉砕は、重合工程の途中及び/又は後に行われ、より好ましくは重合工程後の含水ゲル状重合体に対して行われる。ニーダー重合等、重合中にゲル粉砕を行う形態の場合、単量体水溶液は重合時間の経過とともに連続的に含水ゲル状重合体に変化していく。それゆえ、重合温度が最大となった時点以降の含水ゲル状重合体、或いは単量体の重合率が90モル%以上の含水ゲル状重合体をゲル粉砕すればよい。ここで、重合温度の最大値は、重合ピーク温度とも称される。単量体の重合率は、転化率と称されることもある。単量体の重合率は、含水ゲル状重合体のpH滴定から算出されるポリマー量と、残存モノマー量とから算出される。
乾燥工程は、粒子状含水ゲルを、所望する固形分率まで乾燥させることで、粒状乾燥物を得る工程である。乾燥方法としては、特に限定されないが、例えば、加熱乾燥、熱風乾燥、減圧乾燥、流動層乾燥、赤外線乾燥、マイクロ波乾燥、ドラムドライヤー乾燥、疎水性有機溶媒との共沸脱水による乾燥、高温の水蒸気を利用した高湿乾燥等が挙げられる。また本乾燥工程で後述する後架橋剤を本乾燥工程で使用し、乾燥工程で後架橋(表面架橋ともいう)された吸水性樹脂粉末を得てもよい。
乾燥工程で使用される乾燥装置としては、特に限定されず、伝熱伝導型乾燥機、輻射伝熱型乾燥機、熱風伝熱型乾燥機、誘電加熱型乾燥機等の1種又は2種以上が適宜選択される。乾燥装置は、バッチ式でもよく、連続式でもよい。また、乾燥装置は、直接加熱式でもよく、間接加熱式でもよい。また、乾燥装置は、材料静置型、材料撹拌型、材料移送型、および熱風搬送型のうちのいずれかの乾燥装置であってもよい。例えば、通気バンド式、通気回路式、通気縦型式、平行流バンド式、通気トンネル式、通気攪拌式、通気回転式、加熱管付き回転式、流動層式、気流式等の伝熱型乾燥機が挙げられる。
乾燥工程での乾燥温度は、80℃以上であり、100℃以上であることが好ましく、120℃以上であることがより好ましく、150℃以上であることが特に好ましい。また、乾燥温度は、250℃以下であり、230℃以下であることが好ましく、220℃以下であることがより好ましい。乾燥温度の上限値と下限値はどのような組み合わせであっても好ましい。乾燥温度が80℃未満であると、好適な樹脂固形分(含水率)が得られるまでの乾燥時間が長くなるため好ましくない。また、未乾燥物が生成し、後の粉砕工程時に詰まりが生じ得る。乾燥温度が250℃を超えると、安全面及び着色異物の発生の問題があり好ましくない。なお、乾燥温度とは、直接加熱の場合には乾燥に用いる熱媒の温度を指し、熱風乾燥の場合には乾燥に用いる熱風の温度を指し、間接加熱の場合には乾燥に用いる伝熱面の温度を指す。
乾燥工程での乾燥時間は、固形分が80重量%以上となるまでの時間を指し、60分間以下が好ましく、40分間以下、30分間以下、25分間以下の順に好ましい。乾燥時間の下限値は、乾燥効率を考慮して1分間程度である。さらに、全乾燥時間は、120分間以下であることが好ましく、100分間以下、80分間以下、60分間以下の順により好ましい。乾燥時間が短いと、未乾燥物が生成し、後の粉砕工程時に詰まりが生じ得る。
上記ゲル粉砕工程で得られた粒子状含水ゲルは、上述した乾燥工程で乾燥され、乾燥重合体とされる。乾燥重合体の乾燥減量(粉末又は粒子1gを180℃で3時間加熱して測定)から求められる樹脂固形分は、好ましくは80重量%以上であり、より好ましくは85~99重量%であり、さらに好ましくは86~98重量%である。
本工程は、重合後の含水ゲル、及びその乾燥物に吸水性樹脂の官能基(特にカルボキシル基)と反応する後架橋剤を添加して架橋反応させる工程である。主として、吸水性樹脂粒子の表面から架橋されるため、表面架橋、又は2次架橋とも呼ばれる。一例として、本工程において、粒状含水ゲル及び/又は粒状乾燥物に後架橋剤を添加して反応させる。本工程は、後架橋剤添加工程と、熱処理工程とを有し、必要に応じて熱処理工程後に冷却工程を有していてもよい。
本工程は、粒状乾燥物、又は後架橋された粒状乾燥物の粒度を調整する工程である。この整粒工程によって、粒子径、又は粒度分布がより積極的に制御された吸水性樹脂粉末が得られる。
吸水性樹脂粉末の製造方法には、上述の工程以外にも、冷却工程、単量体水溶液の調製工程、各種添加剤の添加工程、微粉除去工程及び微粉リサイクル工程を含んでいてもよい。さらに、他の公知の工程を含んでいてもよい。
本開示の他の実施形態について、以下に説明する。なお、説明の便宜上、上記実施形態にて説明した部材と同じ機能を有する部材については、同じ符号を付記し、その説明を省略する。
例えば、予測装置100は、吸水性樹脂粉末の製造工程に含まれる重合工程の前、重合工程と乾燥工程との間、及び乾燥工程の後の少なくともいずれかにおいて中間生成物の近赤外吸収スペクトルを測定し、上記工程の任意の段階の中間生成物(又は完成した樹脂粉末)の物性に関連する予測情報を出力してもよい。
実施形態1及び実施形態2に記載の予測装置100の制御ブロック(特に制御部10)の制御ブロック(特に制御部10)は、集積回路(ICチップ)等に形成された論理回路(ハードウェア)によって実現してもよいし、ソフトウェアによって実現してもよい。
〔重合工程〕
内径50mm、容量120mLのポリプロピレン製容器にアクリル酸23.2g、ポリエチレングリコールジアクリレート(重量平均分子量(Mw)523Da)0.135g(0.080モル%)、2.0重量%のジエチレントリアミン5酢酸・3ナトリウム水溶液0.071g、イオン交換水22.2g及び48.5重量%の水酸化ナトリウム水溶液9.6gを混合し、溶液(A)を調製した。
得られた含水ゲル(1)を、以下の仕様を有するスクリュー押出機(ミートチョッパー)でゲル粉砕した。上記スクリュー押出機は、その先端部に多孔板を備え、該多孔板の直径は82mm、孔径8.0mm、孔数33個、厚み9.5mmであった。またゲル粉砕の条件として、含水ゲル(1)の投入量は約360g/分、ゲル投入と並行して90℃の脱イオン水を50g/分で添加しながら、ゲル粉砕を行った。このゲル粉砕された粒子状含水ゲル(1)を後述するゲルD50の評価に用いた。
上記ゲル粉砕された粒子状含水ゲル(1)を目開き850μmのステンレス製金網上に広げ、190℃で30分間熱風乾燥を行った。続いて、該乾燥操作で得られた乾燥重合体(1)をロールミル(有限会社井ノ口技研社製、WML型ロール粉砕機)で粉砕した後、目開き710μm及び目開き175μmのJIS標準篩を用いて分級し吸水性樹脂粉末(1)を得た。
上記吸水性樹脂粉末(1)100gに、エチレングリコールジグリシジルエーテル0.025g、エチレンカーボネート0.3g、プロピレングリコール0.5g及び脱イオン水2.0gからなる表面架橋剤溶液を噴霧して混合した。この混合物を200℃で35分間加熱処理することにより、表面架橋された吸水性樹脂粉末(2)を得た。
近赤外吸収スペクトルの測定機器及び条件は下記の通りである。
測定波長:800~2500nm
測定方式:拡散反射測定
(ii)機器:IRMA5184S(株式会社チノー製)
測定波長(8波長):1320、1460、1600、1720、1800、1960、2100、2310nm
測定方式:近赤外吸収式。
得られた近赤外吸収スペクトルの波長データを特徴量とし、測定した試料の物性情報を目的変数として、PCR又はPLS(Partial least square)回帰分析法により、これらの関係式を求めた。以下に吸水性樹脂粉末の物性(1)ゲルD50、(2)CRC、(3)AAP、(4)SFC、(5)D50及び(6)固形分それぞれについて、予測装置の性能を評価した。
データセットは近赤外測定吸収スペクトルと、該近赤外吸収スペクトルに対応付けられている物性との組み合わせを36個有する。データセットを、学習データと検証データとにランダム分割して、80%を学習用、20%を検証用に用いた。学習データに対してPCRによる予測モデルを作成した。図10は、ゲルD50の実測値に対する予測値を、80~190μmの範囲においてプロットしたグラフである。
データセットは近赤外測定吸収スペクトルと、該近赤外吸収スペクトルに対応付けられている物性との組み合わせを79個有する。データセットを、学習データと検証データとにランダム分割して、80%を学習用、20%を検証用に用いた。学習データに対してPLSによる予測モデルを作成した。図11は、CRCの実測値に対する予測値を、24~31g/gの範囲においてプロットしたグラフである。
データセットは近赤外測定吸収スペクトルと、該近赤外吸収スペクトルに対応付けられている物性との組み合わせを69個有する。データセットを、学習データと検証データとにランダム分割して、80%を学習用、20%を検証用に用いた。学習データに対してPLSによる予測モデルを作成した。図12は、AAPの実測値に対する予測値を、24.5~27g/gの範囲においてプロットしたグラフである。
データセットは近赤外測定吸収スペクトルと、該近赤外吸収スペクトルに対応付けられている物性との組み合わせを64個有する。データセットを、学習データと検証データとにランダム分割して、80%を学習用、20%を検証用に用いた。学習データに対してPLSによる予測モデルを作成した。図13は、SFCの実測値に対する予測値を、20~110(×10-7・cm3・s・g-1)の範囲においてプロットしたグラフである。
データセットは近赤外測定吸収スペクトルと、該近赤外吸収スペクトルに対応付けられている物性との組み合わせを90個有する。データセットを、学習データと検証データとにランダム分割して、80%を学習用、20%を検証用に用いた。学習データに対してPLSによる予測モデルを作成した。図14は、D50の実測値に対する予測値を、250~450μmの範囲においてプロットしたグラフである。
データセットは近赤外測定吸収スペクトルと、該近赤外吸収スペクトルに対応付けられている物性との組み合わせを29個有する。データセットを、学習データと検証データとにランダム分割して、80%を学習用、20%を検証用に用いた。学習データに対してPLSによる予測モデルを作成した。図15は、含水率の実測値に対する予測値を96.5~98.5wt%の範囲においてプロットしたグラフである。固形分は、100-含水率(重量%)によって求められるため、本グラフは、固形分率における実測値に対する予測値を示すものともと言える。
いずれの物性においても、予測値は、実測値とよい相関を示した。また、学習データに含まれていない検証データについても、学習データと同様の精度で各物性が予測できており、予測装置100が良好な性能を有することが示された。
11 測定データ取得部
13 予測部
22 予測モデル
23 物性情報
Claims (11)
- 樹脂粉末の物性を予測する予測方法であって、
前記樹脂粉末は、吸水性樹脂粉末、及び該吸水性樹脂粉末を製造するための製造工程において生成される中間生成物のいずれかであり、
前記樹脂粉末の近赤外吸収スペクトルを示す近赤外測定データを取得する近赤外測定データ取得ステップと、
前記近赤外測定データ、及び、前記近赤外測定データに基づいて生成された1以上の加工データ、の少なくとも1つ以上を予測モデルに入力して、該樹脂粉末の物性に関連する予測情報を出力する予測ステップと、を含む、
予測方法。 - 前記予測モデルは、(1)過去に製造された、物性が既知の複数の製造済樹脂粉末の近赤外吸収スペクトルを含む近赤外測定データと、該近赤外測定データに対応付けられている最終生成物の物性情報との組み合わせ、及び(2)各製造済樹脂粉末を製造するための製造工程において生成された、物性が既知の複数の生成済中間生成物の近赤外吸収スペクトルを含む近赤外測定データと、該近赤外測定データに対応付けられている中間生成物の物性情報との組み合わせ、のうち少なくともいずれかを学習データとして用いた機械学習によって生成されたものである、
請求項1に記載の予測方法。 - 前記予測モデルは、線形回帰、及び非線形回帰のうちのいずれかを用いて生成されたものである、
請求項2に記載の予測方法。 - 前記予測モデルは、主成分回帰、及び部分的最小二乗回帰のうちのいずれかを用いて生成されたものである、
請求項2又は3に記載の予測方法。 - 前記加工データを生成する前処理ステップを含み、
前記前処理ステップにおいて、外れ値除去処理、平均化処理、波長範囲選択処理、及び微分処理のうちのいずれか1以上を行う、
請求項1から4のいずれか1項に記載の予測方法。 - 前記予測情報は、(1)前記中間生成物である含水ゲルの質量平均粒子径(ゲルD50)、(2)前記樹脂粉末の無加圧下吸収倍率(CRC)、(3)前記樹脂粉末の加圧下吸収倍率(AAP)、(4)前記樹脂粉末の食塩水流れ誘導性(SFC)、(5)前記樹脂粉末の質量平均粒子径(D50)、及び(6)前記樹脂粉末の固形分含有量又は固形分率、のうち少なくともいずれか1つを含む、
請求項1から5のいずれか1項に記載の予測方法。 - 前記樹脂粉末の前記製造工程は、重合工程、乾燥工程を含み、
前記近赤外吸収スペクトルは、前記重合工程の前、前記重合工程と前記乾燥工程との間、及び前記乾燥工程の後の少なくともいずれかにおいて測定され、
前記予測ステップにおいて出力される前記予測情報に基づいて、前記樹脂粉末の製造工程で使用されるいずれか1以上の製造装置が制御される、
請求項1から6のいずれか1項に記載の予測方法。 - 樹脂粉末の物性を予測する予測装置であって、
前記樹脂粉末は、吸水性樹脂粉末、及び該吸水性樹脂粉末を製造するための製造工程において生成される中間生成物のいずれかであり、
前記樹脂粉末について測定された近赤外吸収スペクトルを示す近赤外測定データを取得する測定データ取得部と、
前記近赤外測定データ、及び、前記近赤外測定データに基づいて生成された1以上の加工データ、の少なくともいずれかを予測モデルに入力して、該樹脂粉末の物性に関連する予測情報を出力する予測部と、を備える、
予測装置。 - 重合工程及び乾燥工程を含む樹脂粉末の製造方法であって、
請求項1から7のいずれか1項に記載の予測方法により得られた予測情報に基づき、
前記樹脂粉末のいずれか1以上の製造工程において、その製造条件が制御される、樹脂粉末の製造方法。 - 樹脂粉末の製造方法を制御するための、請求項1から7のいずれか1項に記載の予測方法により得られた予測情報の使用。
- 請求項1から6のいずれか1項に記載の予測方法に用いる樹脂粉末の近赤外吸収スペクトルを測定する測定方法であって、
前記樹脂粉末に近赤外線を照射するステップと、
前記樹脂粉末からの反射光及び透過光のうちの少なくとも一方を測定した測定値から前記樹脂粉末の近赤外吸収スペクトルを算出するステップと、を含み、
前記樹脂粉末は、吸水性樹脂粉末、及び該吸水性樹脂粉末を製造するための製造工程において生成される中間生成物のいずれかである、
測定方法。
Priority Applications (5)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US18/031,241 US20230375472A1 (en) | 2020-10-13 | 2021-10-12 | Prediction method and prediction device |
| EP21880105.8A EP4231000A4 (en) | 2020-10-13 | 2021-10-12 | PREDICTION METHOD AND PREDICTION DEVICE |
| KR1020237016195A KR102917557B1 (ko) | 2020-10-13 | 2021-10-12 | 예측 방법, 예측 장치 |
| JP2022557007A JPWO2022080367A1 (ja) | 2020-10-13 | 2021-10-12 | |
| CN202180070148.0A CN116368370A (zh) | 2020-10-13 | 2021-10-12 | 预测方法、预测装置 |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2020-172722 | 2020-10-13 | ||
| JP2020172722 | 2020-10-13 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2022080367A1 true WO2022080367A1 (ja) | 2022-04-21 |
Family
ID=81208044
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/JP2021/037735 Ceased WO2022080367A1 (ja) | 2020-10-13 | 2021-10-12 | 予測方法、予測装置 |
Country Status (6)
| Country | Link |
|---|---|
| US (1) | US20230375472A1 (ja) |
| EP (1) | EP4231000A4 (ja) |
| JP (1) | JPWO2022080367A1 (ja) |
| KR (1) | KR102917557B1 (ja) |
| CN (1) | CN116368370A (ja) |
| WO (1) | WO2022080367A1 (ja) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2024003699A (ja) * | 2022-06-27 | 2024-01-15 | Dic株式会社 | 樹脂組成物の製造条件の判定方法 |
| WO2024062093A1 (en) * | 2022-09-23 | 2024-03-28 | Basf Se | Apparatus for determining a technical application property of a superabsorbent material |
| WO2024202794A1 (ja) * | 2023-03-30 | 2024-10-03 | 株式会社日立ハイテク | スペクトル解析装置、スペクトル解析システムおよびスペクトル解析方法 |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119198627B (zh) * | 2024-11-07 | 2025-02-21 | 泰和新材集团股份有限公司 | 一种间位芳纶树脂pH值快速在线检测方法 |
| CN120011708B (zh) * | 2025-02-19 | 2025-10-24 | 湖南中烟工业有限责任公司 | 一种近红外温度校正模型构建方法、装置、设备及介质 |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5669894A (en) | 1994-03-29 | 1997-09-23 | The Procter & Gamble Company | Absorbent members for body fluids having good wet integrity and relatively high concentrations of hydrogel-forming absorbent polymer |
| JPH09268232A (ja) * | 1996-02-02 | 1997-10-14 | Nippon Shokubai Co Ltd | 親水性重合体の変性方法並びに親水性樹脂の製造方法 |
| JP2000063527A (ja) | 1998-08-12 | 2000-02-29 | Nippon Shokubai Co Ltd | 含水ゲル状架橋重合体の細粒化方法 |
| JP2002090298A (ja) * | 2000-09-01 | 2002-03-27 | Spectron Tech Co Ltd | 近赤外線分光分析法による皮膚水分測定方法および装置 |
| JP2003344279A (ja) * | 2002-05-29 | 2003-12-03 | Shiseido Co Ltd | 毛髪水分の測定方法 |
| US7265190B2 (en) | 2002-11-07 | 2007-09-04 | Nippon Shokubai Co., Ltd. | Process and apparatus for production of water-absorbent resin |
| WO2009016055A2 (en) | 2007-07-27 | 2009-02-05 | Basf Se | Water-absorbing polymeric particles and method for the production thereof |
| US7638570B2 (en) | 2003-02-10 | 2009-12-29 | Nippon Shokubai Co., Ltd. | Water-absorbing agent |
| WO2012043821A1 (ja) * | 2010-09-30 | 2012-04-05 | 株式会社日本触媒 | 粒子状吸水剤及びその製造方法 |
| US20120318046A1 (en) | 2011-06-17 | 2012-12-20 | The Procter & Gamble Company | Method For Determining Properties Of Superabsorbent Polymer Particles And Of Absorbent Structures Containing Such Particles |
| JP2013018917A (ja) * | 2011-07-13 | 2013-01-31 | Kao Corp | ポリウロン酸又はその塩の製造方法 |
| WO2016204302A1 (ja) | 2015-06-19 | 2016-12-22 | 株式会社日本触媒 | ポリ(メタ)アクリル酸(塩)系粒子状吸水剤及び製造方法 |
| WO2020109601A1 (en) | 2018-11-29 | 2020-06-04 | Basf Se | Prediction of physical properties of superabsorbent polymers |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO1996007092A1 (en) * | 1994-09-01 | 1996-03-07 | E.I. Du Pont De Nemours And Company | Process for neutralizing acids in a polymer solution |
| JPH11315137A (ja) * | 1998-04-30 | 1999-11-16 | Mitsubishi Chemical Corp | ポリエステル樹脂の製造方法 |
| CN1191466C (zh) * | 2002-07-30 | 2005-03-02 | 中国石油化工股份有限公司 | 测定聚丙烯树脂等规指数的方法 |
| ITMI20131916A1 (it) * | 2013-11-19 | 2015-05-20 | Versalis Spa | Metodo per monitorare un parametro di controllo di una reazione di polimerizzazione e relativo apparato per attuare detto metodo |
| PL2915658T3 (pl) * | 2014-03-05 | 2018-01-31 | Flooring Technologies Ltd | Sposób określania zawartości wilgoci warstwy żywicy na płycie z tworzywa drzewnego pokrytej materiałem podłożowym zawierającym włókna |
| US9389161B2 (en) * | 2014-04-09 | 2016-07-12 | Exxonmobil Chemical Patents Inc. | On-line FT-NIR method to determine particle size and distribution |
| JP6438549B1 (ja) * | 2017-08-24 | 2018-12-12 | 日本分光株式会社 | 機械学習を用いた未知化合物の分類方法 |
| EP3865872B1 (en) * | 2018-11-19 | 2025-01-08 | Canon Kabushiki Kaisha | Information processing device, information processing device control method, program, calculation device, and calculation method |
-
2021
- 2021-10-12 JP JP2022557007A patent/JPWO2022080367A1/ja active Pending
- 2021-10-12 KR KR1020237016195A patent/KR102917557B1/ko active Active
- 2021-10-12 US US18/031,241 patent/US20230375472A1/en active Pending
- 2021-10-12 CN CN202180070148.0A patent/CN116368370A/zh active Pending
- 2021-10-12 EP EP21880105.8A patent/EP4231000A4/en active Pending
- 2021-10-12 WO PCT/JP2021/037735 patent/WO2022080367A1/ja not_active Ceased
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5669894A (en) | 1994-03-29 | 1997-09-23 | The Procter & Gamble Company | Absorbent members for body fluids having good wet integrity and relatively high concentrations of hydrogel-forming absorbent polymer |
| JPH09268232A (ja) * | 1996-02-02 | 1997-10-14 | Nippon Shokubai Co Ltd | 親水性重合体の変性方法並びに親水性樹脂の製造方法 |
| JP2000063527A (ja) | 1998-08-12 | 2000-02-29 | Nippon Shokubai Co Ltd | 含水ゲル状架橋重合体の細粒化方法 |
| JP2002090298A (ja) * | 2000-09-01 | 2002-03-27 | Spectron Tech Co Ltd | 近赤外線分光分析法による皮膚水分測定方法および装置 |
| JP2003344279A (ja) * | 2002-05-29 | 2003-12-03 | Shiseido Co Ltd | 毛髪水分の測定方法 |
| US7265190B2 (en) | 2002-11-07 | 2007-09-04 | Nippon Shokubai Co., Ltd. | Process and apparatus for production of water-absorbent resin |
| US7638570B2 (en) | 2003-02-10 | 2009-12-29 | Nippon Shokubai Co., Ltd. | Water-absorbing agent |
| WO2009016055A2 (en) | 2007-07-27 | 2009-02-05 | Basf Se | Water-absorbing polymeric particles and method for the production thereof |
| WO2012043821A1 (ja) * | 2010-09-30 | 2012-04-05 | 株式会社日本触媒 | 粒子状吸水剤及びその製造方法 |
| US20120318046A1 (en) | 2011-06-17 | 2012-12-20 | The Procter & Gamble Company | Method For Determining Properties Of Superabsorbent Polymer Particles And Of Absorbent Structures Containing Such Particles |
| JP2013018917A (ja) * | 2011-07-13 | 2013-01-31 | Kao Corp | ポリウロン酸又はその塩の製造方法 |
| WO2016204302A1 (ja) | 2015-06-19 | 2016-12-22 | 株式会社日本触媒 | ポリ(メタ)アクリル酸(塩)系粒子状吸水剤及び製造方法 |
| WO2020109601A1 (en) | 2018-11-29 | 2020-06-04 | Basf Se | Prediction of physical properties of superabsorbent polymers |
Non-Patent Citations (2)
| Title |
|---|
| MATSUDA, MASAAKI ET AL.: "Hydrolysis of polyacrylonitrile by ammonium hydroxide; water absorbing performance of the hydrolyzed polymer", POLYMER PREPRINTS, JAPAN, vol. 38, no. 2, 12 September 1989 (1989-09-12), pages 1 - 2, XP009536016 * |
| See also references of EP4231000A4 |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2024003699A (ja) * | 2022-06-27 | 2024-01-15 | Dic株式会社 | 樹脂組成物の製造条件の判定方法 |
| JP7829420B2 (ja) | 2022-06-27 | 2026-03-13 | Dic株式会社 | 樹脂組成物の製造条件の判定方法 |
| WO2024062093A1 (en) * | 2022-09-23 | 2024-03-28 | Basf Se | Apparatus for determining a technical application property of a superabsorbent material |
| WO2024202794A1 (ja) * | 2023-03-30 | 2024-10-03 | 株式会社日立ハイテク | スペクトル解析装置、スペクトル解析システムおよびスペクトル解析方法 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116368370A (zh) | 2023-06-30 |
| US20230375472A1 (en) | 2023-11-23 |
| KR20230086762A (ko) | 2023-06-15 |
| EP4231000A4 (en) | 2024-11-20 |
| JPWO2022080367A1 (ja) | 2022-04-21 |
| EP4231000A1 (en) | 2023-08-23 |
| KR102917557B1 (ko) | 2026-01-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2022080367A1 (ja) | 予測方法、予測装置 | |
| JPWO2022080367A5 (ja) | ||
| US11020726B2 (en) | Super absorbent polyacrylic acid (salt)-based resin powder, method for manufacturing same, and method for evaluating same | |
| US9447203B2 (en) | Method for producing water absorbent polyacrylic acid (salt) resin powder, and water absorbent polyacrylic acid (salt) resin powder | |
| KR102297640B1 (ko) | 겔 분쇄 장치, 폴리아크릴산(염)계 흡수성 수지 분말의 제조 방법, 및 흡수성 수지 분말 | |
| US12123835B2 (en) | Prediction of physical properties of superabsorbent polymers | |
| CN100471874C (zh) | 超吸收剂、其制备方法及其用途 | |
| US8975339B2 (en) | Surface treatment method for water-absorbent resin | |
| CN108350179B (zh) | 超吸收性聚合物 | |
| CA2649299A1 (en) | An absorbent member comprising a modified water absorbent resin | |
| WO2011090130A1 (ja) | 吸水性樹脂の製造方法 | |
| Pourjavadi et al. | Carrageenan‐g‐poly (acrylamide)/poly (vinylsulfonic acid, sodium salt) as a novel semi‐IPN hydrogel: Synthesis, characterization, and swelling behavior | |
| KR102801861B1 (ko) | 흡수체, 흡수성 수지 및 흡수성 물품 | |
| JP2016112475A (ja) | ポリ(メタ)アクリル酸塩系吸水性樹脂を主成分とする吸水剤及びその製造方法 | |
| JP7187220B2 (ja) | 吸収性樹脂粒子及びその製造方法 | |
| JP2016113465A (ja) | ポリアクリル酸(塩)系吸水性樹脂及びその製造方法 | |
| KR20250076538A (ko) | 초흡수성 재료의 기술적 응용 특성을 결정하기 위한 장치 | |
| WO2019188648A1 (ja) | 吸水性樹脂粒子及びその製造方法 | |
| WO2023149574A1 (ja) | 吸水性樹脂の重合方法 | |
| CN118055954A (zh) | 制备超级吸收剂颗粒的方法 | |
| JPH01131209A (ja) | 含水ゲル状重合体の中和方法 | |
| JP2018065905A (ja) | ポリアクリル酸(塩)系吸水性樹脂粉末の製造方法 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21880105 Country of ref document: EP Kind code of ref document: A1 |
|
| ENP | Entry into the national phase |
Ref document number: 2022557007 Country of ref document: JP Kind code of ref document: A |
|
| ENP | Entry into the national phase |
Ref document number: 20237016195 Country of ref document: KR Kind code of ref document: A |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| ENP | Entry into the national phase |
Ref document number: 2021880105 Country of ref document: EP Effective date: 20230515 |