JP7501432B2 - Waste information prediction device, incinerator combustion control device, waste information prediction method, waste information prediction model learning method, and waste information prediction model program - Google Patents

Description

The present invention relates to a waste information prediction device, an incinerator combustion control device, a waste information prediction method, a waste information prediction model learning method, and a waste information prediction model program.

In waste treatment plants such as incinerators, knowing information about the waste being fed into the plant (such as calories) in advance is extremely effective in ensuring stable combustion of the waste inside the furnace.

For example, Patent Document 1 discloses a technology for predicting information related to waste, which uses optical flow to match time-series images of the burning state of waste and learns a prediction model that predicts information related to the waste. Patent Document 2 discloses a technology for detecting the point at which waste starts to burn in the dry zone, based on a two-dimensional image that shows the burning temperature distribution of waste closer to the dry zone and records only wavelengths in the 3-4 μm band.

JP 2020-119407 A Japanese Patent Application Laid-Open No. 8-49830 Japanese Patent Application Laid-Open No. 6-174219

However, the technologies disclosed in Patent Documents 1 and 2 do not extract features that are correlated with the state of waste in an incinerator, making it difficult to accurately predict information about the waste. Furthermore, even if such low-precision features were used, it was not possible to efficiently control waste treatment plants such as incinerators.

Patent Document 3 also discloses a technology for estimating the moisture content of waste based on the temperature distribution in the thickness direction cross section of the waste before the waste is transported to the combustion zone. However, Patent Document 3 requires that the temperature distribution in the thickness direction cross section be grasped, which requires temperature information for a relatively wide area of the waste. As a result, it takes time to grasp the temperature distribution, and the temperature distribution depends on the residence time of the waste in the furnace before it falls, so the accuracy of grasping the waste quality is not very high.

The present invention has been made in consideration of the above, and aims to provide a waste information prediction device, an incinerator combustion control device, a waste information prediction method, a waste information prediction model learning method, and a waste information prediction model program that can accurately predict information about waste in an incinerator.

In order to solve the above-mentioned problems and achieve the object, the waste information prediction device of the present invention is a waste information prediction device that predicts information about waste in an incinerator, and has an input unit that can input photographed image information of the waste photographed outside, a feature calculation means that divides the photographed image information of the waste input from the input unit into multiple blocks and tracks at least one of the divided multiple blocks for a predetermined period of time to calculate the feature amount of the tracked block, a storage unit that stores a relationship model that associates the feature amount of the waste with information about the waste, and a control means, and the control means calculates the feature amount of the waste by the feature calculation means from the photographed image information obtained by photographing the waste in the incinerator input from the input unit, and predicts information about the waste based on the relationship model stored in the storage unit.

The waste information prediction device according to the present invention is characterized in that, in the above invention, the feature amount includes at least one of the amount of temperature change of the waste, the amount of movement of the waste, and the total number of multiple blocks into which the waste is divided and set in the captured image information.

The waste information prediction device according to the present invention is characterized in that, in the above invention, the information about the waste includes at least one of the calories generated when the waste is burned and the amount of water vapor generated from the boiler of the incinerator when the waste is burned.

The waste information prediction device according to the present invention is characterized in that it has an output unit that outputs information about the waste predicted by the control means in the above invention.

The waste information prediction device according to the present invention is characterized in that, in the above invention, the photographed image information of the waste input from the input unit is photographed image information obtained by photographing the waste before it is supplied to the grate of the incinerator.

The waste information prediction device according to the present invention is characterized in that, in the above invention, the relational model stored in the storage unit is a learned waste information prediction model obtained by learning from the relationship between the feature quantities of past waste and information about the past waste.

The waste information prediction device according to the present invention is characterized in that, in the above invention, the relational model stored in the storage unit is a lookup table or a predetermined function in which the feature quantities of past waste are input values and information about the past waste is output values.

In order to solve the above-mentioned problems and achieve the objectives, the incinerator combustion control device of the present invention is a combustion control device that includes a combustion control unit that controls an incinerator that combusts waste, and is characterized in that the combustion control unit controls combustion in the incinerator based on information about waste output from the waste information prediction device.

The incinerator combustion control device according to the present invention is characterized in that, in the above invention, the incinerator comprises a grate for moving the waste, a waste supply device for supplying the waste onto the grate, and a blower for blowing air into the incinerator, and the combustion control unit controls at least one of the speed at which the waste moves on the grate, the speed at which the waste is supplied by the waste supply device, the amount of air blown by the blower, and the temperature of the air blown by the blower based on information about the waste.

In order to solve the above-mentioned problems and achieve the objective, the waste information prediction method of the present invention is a waste information prediction method for predicting information about waste in an incinerator, and is characterized by having an image acquisition process for acquiring photographed image information of the waste put into the incinerator, a feature amount calculation process for dividing the photographed image information of the waste acquired in the image acquisition process into a plurality of blocks, tracking at least one of the divided blocks and calculating the feature amount of the tracked block, and a waste information prediction process for predicting information about the waste from the feature amount of the waste calculated in the feature amount calculation process.

The waste information prediction method according to the present invention is characterized in that, in the above invention, the waste information prediction step makes predictions using a relational model in which the waste characteristics and information about the waste are previously associated with each other.

The waste information prediction method according to the present invention is characterized in that, in the above invention, the relational model is a learned waste information prediction model obtained by learning from the relationship between the feature quantities of past waste and information about the past waste.

The waste information prediction method according to the present invention is characterized in that, in the above invention, the relational model is a lookup table or a predetermined function that uses the feature quantities of past waste as input values and information about the past waste as output values.

The waste information prediction method according to the present invention is characterized in that, in the above invention, the photographed image information of the waste acquired in the image acquisition process is photographed image information obtained by photographing the waste before it is supplied to the grate of the incinerator.

The waste information prediction method according to the present invention is characterized in that it further includes a combustion control step for controlling combustion in the incinerator based on the waste information predicted in the waste information prediction step.

In order to solve the above-mentioned problems and achieve the objective, the waste information prediction model learning method of the present invention is a waste information prediction model learning method for predicting information about waste in an incinerator, which is characterized in that it divides photographed image information of waste in an incinerator that incinerates waste into a plurality of blocks, tracks at least one of the plurality of divided blocks, uses a feature amount calculated from the tracked block as an input value, and uses information about the waste as an output value, and causes the waste information prediction model to learn the relationship between the past feature amount and the corresponding past information about the waste.

In order to solve the above-mentioned problems and achieve the objective, the waste information prediction model program of the present invention is a waste information prediction model program that can be recorded on a recording medium and predicts information about waste in an incinerator, and is characterized in that it is trained to output information about the waste when a feature calculated from a tracking block that tracks multiple blocks set in photographed image information of the waste is input.

According to the waste information prediction device, incinerator combustion control device, waste information prediction method, waste information prediction model learning method, and waste information prediction model program of the present invention, the movement of waste within the incinerator is tracked in multiple pre-set blocks, and the feature values of each block are trained into a prediction model, making it possible to accurately predict information about waste within the incinerator, and to efficiently control waste treatment plants such as incinerators based on the predicted waste information.

FIG. 1 is a diagram showing a schematic diagram of the overall configuration of an incinerator to which a waste information prediction device according to an embodiment of the present invention is applied. FIG. 2 is a side view showing waste, a portion where the waste is fed onto a fire grate, and an imaging unit in an incinerator according to an embodiment of the present invention. FIG. 3 is a front view showing the field of view of an imaging unit in the incinerator according to the embodiment of the present invention. FIG. 4 is a block diagram showing the configuration of an incinerator combustion control device and a waste information prediction device according to an embodiment of the present invention. FIG. 5 is a diagram showing an example of photographed image information captured by an imaging unit in the incinerator according to the embodiment of the present invention. FIG. 6 is a graph summarizing the transition of the average temperature of each area of the photographed image information, the transition of the total average temperature of each area, and the transition of the calories of the pre-supply waste related to the photographed image information. FIG. 7 is a diagram showing an example in which optical flow based on a block matching method is applied to actual photographed image information captured by an imaging unit in an incinerator according to an embodiment of the present invention. FIG. 8 is a diagram showing an example in which optical flow based on a block matching method is applied to actual photographed image information captured by an imaging unit in an incinerator according to an embodiment of the present invention. FIG. 9 is a diagram showing an example in which optical flow based on a block matching method is applied to actual photographed image information captured by an imaging unit in an incinerator according to an embodiment of the present invention. FIG. 10 is a diagram for explaining a method for calculating the average temperature rise amount per block as a feature amount in the waste information prediction device according to the embodiment of the present invention. FIG. 11 is a diagram for explaining a method for calculating the amount of waste movement for each block as a feature amount in the waste information prediction device according to the embodiment of the present invention. FIG. 12 is a flowchart showing the processing steps of the learning method for the waste information prediction model according to the embodiment of the present invention. FIG. 13 is a flowchart showing the processing steps of the waste information prediction method according to the embodiment of the present invention.

The waste information prediction device, incinerator combustion control device, waste information prediction method, waste information prediction model learning method, and waste information prediction model program according to the embodiments of the present invention will be described with reference to the drawings. Note that in the drawings referred to in the following embodiments, the same or corresponding parts are given the same reference numerals. Furthermore, the present invention is not limited to the following embodiments.

Here, in order to efficiently control, for example, a grate (stoker) type incinerator, it is important to understand information about the waste being burned in the incinerator (hereinafter referred to as "waste information"). Therefore, the present inventor tracks the movement of waste inside the incinerator in multiple pre-set blocks, trains a prediction model to learn the features of each block, and predicts (estimates) waste information using the prediction model. The embodiment described below was devised based on the inventor's diligent investigations.

(Incinerator)
An incinerator to which the waste information prediction device according to the embodiment of the present invention is applied will be described with reference to Fig. 1. The incinerator is, for example, a grate-type incinerator, and is installed in a plant that generates power using waste. The incinerator includes a furnace 1 in which waste is burned, a waste inlet 2 for inputting waste, a waste supply device 3, a grate 4, an ash drop port 5, a blower (combustion air blower 6, secondary air blower 11) for blowing air into the furnace 1, a furnace outlet 7, a chimney 8, a boiler 9 equipped with a heat exchanger 9a and a steam drum 9b, and a secondary air inlet 10.

A combustion air damper 14 is provided immediately adjacent to the combustion air blower 6. In addition, under-grate combustion air dampers 14a, 14b, 14c, and 14d are provided in the wind box below the grate 4. In addition, a secondary air damper 15 is provided immediately adjacent to the secondary air blower 11. In addition, an intermediate ceiling 16 is provided in the furnace 1 at an upper position along the height direction of the furnace 1.

In addition, the furnace 1 is provided with a combustion chamber gas thermometer 17, a main flue gas thermometer 18, a furnace outlet lower gas thermometer 19, a furnace outlet middle gas thermometer 20, and a furnace outlet gas thermometer 21. The boiler 9 is provided with a boiler outlet oxygen concentration meter 22 and a steam flowmeter 25. The inlet of the chimney 8 is provided with a gas concentration meter 23. The piping connecting the outlet of the boiler 9 and the chimney 8 is provided with an exhaust gas flowmeter 24. The measurements taken by these instruments are sent to the combustion control device 30 as combustion process measurements and are stored in the memory unit 32 (see FIG. 4).

An imaging unit 26 is provided downstream in the waste transport direction in the furnace 1. The imaging unit 26 is configured with a flame-transmitting camera, for example an infrared camera, and an image processing unit that processes the captured image data.

The installation state of the imaging unit 26 will be described with reference to FIG. 2. As shown in the figure, the imaging unit 26 is disposed outside the furnace 1, close to a monitoring window provided in the furnace wall 1a. The imaging unit 26 may be disposed inside the furnace 1, for example, by adding a water-cooling structure. The waste 50 falls from the waste supply unit 12 onto the grate 4 at the step wall 13. The waste 50 that has fallen onto the grate 4 is stirred by the reciprocating motion that accompanies the back and forth movement of the grate 4, and moves forward toward the imaging unit 26.

The imaging unit 26 acquires thermographic information of the waste 50 on the grate 4 (hereinafter referred to as "waste on the grate 52") as photographed image information (thermal image information). Here, the wavelength of the infrared rays emitted from the waste 50 is different from the wavelength of the infrared rays emitted from the high-temperature gas and flame in the space. Therefore, in the imaging unit 26, by appropriately selecting the infrared wavelength to be measured, it is possible to obtain photographed image information corresponding to the temperature distribution of the layer of the waste on the grate 52 even if a flame is present within the measurement field of view. In addition, the imaging unit 26 can obtain photographed image information of the layer of the waste on the grate 52 at a position upstream of the combustion area (upstream of the flame) by setting the measurement range in the furnace length direction. The photographed image information can be treated as video data in a state where the flame is transmitted, that is, as multiple image data.

In other words, the imaging unit 26 can capture images of the waste 50 (hereinafter referred to as "waste before supply 51") sent out from the waste supply unit 12, the step wall 13 having a step onto which the waste 50 falls, the waste on the grate 52, and the upper surface of the grate 4, all in a state where the flame is transmitted through the image. The incinerator may further include a combustion image capturing unit that captures the combustion state of the waste on the grate 52, i.e., the flame itself. The captured image information captured by the imaging unit 26 in a state where the flame is transmitted through the image is transmitted to the waste information prediction device 40 immediately or at a predetermined time interval. The captured image information captured by the imaging unit 26 may be stored in the memory unit 32 of the combustion control device 30 and then transmitted from the combustion control device 30 to the waste information prediction device 40.

The imaging unit 26 is installed, for example, in a position that is approximately directly opposite the waste supply unit 12 and the step wall 13. Note that the installation of the imaging unit 26 is not limited to a position that is approximately directly opposite the waste supply unit 12 and the step wall 13.

An example of the field of view of the imaging unit 26 will be described with reference to FIG. 3. As shown in the figure, the imaging unit 26 has a measurement field of view that extends, for example, in the vertical direction and the furnace width direction (left and right direction) of the furnace 1. The field of view of the imaging unit 26 includes the waste supply unit 12, the step wall 13, the grate 4, and the left and right furnace walls 1a. The furnace walls 1a included in the field of view of the imaging unit 26 restrict the outward movement, i.e., the spread, of the waste 50 in the left and right direction. It is preferable that the imaging unit 26 is configured so that it can capture an image of the waste 50 transported to the waste supply unit 12. This makes it possible to capture an image of the waste 50 falling at the position of the step wall 13.

The configurations of the combustion control device 30 and the waste information prediction device 40 will be described with reference to FIG. 4. The combustion control device 30 and the waste information prediction device 40 are connected via a network (not shown) consisting of one or a combination of a dedicated line, a public communication network such as the Internet (e.g., a LAN (Local Area Network), a WAN (Wide Area Network), and a telephone communication network such as a mobile phone, a public line, and a VPN (Virtual Private Network)).

The combustion control device 30 and the waste information prediction device 40 may be configured separately as shown in FIG. 4, or may be configured as an integrated device. Furthermore, the combustion control device 30 and the waste information prediction device 40 may be installed in the same facility as the incinerator, or in a facility separate from the incinerator. When the combustion control device 30 and the waste information prediction device 40 are installed in separate facilities, various information and data are communicated via the network described above.

The combustion control device 30 includes a control unit 31, a memory unit 32, and an operation amount adjustment unit 33. The control unit 31 and the operation amount adjustment unit 33 function as a combustion control unit that controls the incinerator. Specifically, the control unit 31 and the operation amount adjustment unit 33 include a processor such as a CPU (Central Processing Unit), DSP (Digital Signal Processor), or FPGA (Field-Programmable Gate Array) having hardware, and a main memory unit such as a RAM (Random Access Memory) or ROM (Read Only Memory) (none of which are shown).

The storage unit 32 is composed of storage media such as volatile memory such as RAM, non-volatile memory such as ROM, EPROM (Erasable Programmable ROM), hard disk drive (HDD, Hard Disk Drive), and removable media. Examples of removable media include USB (Universal Serial Bus) memory, CD (Compact Disc), DVD (Digital Versatile Disc), and BD (Blu-ray (registered trademark) Disc). The storage unit 32 may also be composed of a computer-readable storage medium such as an externally attachable memory card.

The memory unit 32 can store an operating system (OS), various programs, various tables, various databases, and the like for executing the operation of the combustion control device 30. Here, the various programs also include information processing programs that realize processing based on models such as the learning model and the trained model according to this embodiment. These various programs can also be recorded on computer-readable recording media such as hard disks, flash memories, CD-ROMs, DVD-ROMs, and flexible disks, and widely distributed.

The combustion control device 30 performs a combustion control process to control the combustion of the waste 50 in the incinerator as the operation amount of each operation terminal based on the waste information output from the waste information prediction device 40. Specifically, the combustion control device 30 controls at least one of the following based on the waste information: the movement speed of the waste 50 on the grate 4 (also called the "grate feed speed"), the supply speed of the waste 50 by the waste supply device 3 (also called the "waste supply device feed speed"), the amount and temperature of the combustion air blown by the combustion air blower 6, and the amount and temperature of the secondary air blown by the secondary air blower 11.

The combustion control device 30 may control the above-mentioned manipulated variables by taking into consideration a predetermined manipulated variable reference value setting relational expression (hereinafter referred to as the "manipulated variable relational expression") in addition to the waste information acquired from the waste information prediction device 40. This manipulated variable relational expression is, for example, a relational expression between a waste incineration amount set value and a manipulated variable reference value (target value of the manipulated variable), and includes a control parameter as a correction coefficient. The control parameter is adjusted by the control unit 31 so as to match the waste incineration amount set value. When the waste incineration amount set value is changed, the adjusted control parameter is changed by the control unit 31 in accordance with the changed set value. The preset manipulated variable reference value is corrected by changing the control parameter.

The control unit 31 adjusts the operation quantity reference value based on the waste information acquired from the waste information prediction device 40 and the adjustment of the control parameters in the operation quantity relational expression that is considered as necessary. The control unit 31 corrects the adjusted operation quantity reference value based on a predetermined control algorithm such as PID control or fuzzy calculation. The memory unit 32 stores data referenced by the control unit 31. The memory unit 32 stores, for example, a predetermined operation quantity relational expression, a control algorithm, a predetermined waste incineration quantity setting value, and combustion process measurement values acquired as combustion state quantities in the furnace 1.

The operation amount adjustment unit 33 adjusts the operation amount of each operation terminal so that it follows the operation amount reference value. Specifically, the operation amount adjustment unit 33 includes a combustion air adjustment unit 331, an air amount ratio adjustment unit 332, a secondary air adjustment unit 333, a waste supply device feed rate adjustment unit 334, and a grate feed rate adjustment unit 335.

The combustion air adjustment unit 331 adjusts the operation amount so that the combustion air blowing amount and temperature follow the operation amount reference value corrected by the control unit 31 (hereinafter referred to as the "corrected operation amount reference value"). The air amount ratio adjustment unit 332 controls each of the under-grate combustion air dampers 14a to 14d to adjust the relative ratio of the flow rates in each wind box. The secondary air adjustment unit 333 adjusts the operation amount so that the secondary air blowing amount and temperature follow the corrected operation amount reference value. Here, the combustion air and secondary air blowing amounts are adjusted by controlling the opening degree of each of the combustion air damper 14, the under-grate combustion air dampers 14a to 14d, and the secondary air damper 15.

The waste feeder feed speed adjustment unit 334 adjusts the operation amount so that the waste feeder feed speed follows the corrected operation amount reference value. The grate feed speed adjustment unit 335 adjusts the operation amount so that the grate feed speed follows the corrected operation amount reference value. When the operation amount reference value is not corrected by the control unit 31, the operation amount adjustment unit 33 adjusts each operation amount based on the uncorrected operation amount reference value.

The waste information prediction device 40 includes a control unit (control means) 41, an output unit 42, an input unit 43, and a storage unit 44. The control unit 41 has a similar configuration to the control unit 31 described above both functionally and physically, and includes a processor such as a CPU, DSP, FPGA, etc. having hardware, and a main storage unit such as a RAM or ROM (none of which are shown). Specifically, the control unit 41 functions as a feature amount calculation unit (feature amount calculation means) 411, a learning unit 412, and a waste information prediction unit 413.

The feature calculation unit 411 calculates feature quantities characterizing the state of the waste 50 from the captured image information of the waste 50 in the furnace 1. Specifically, the feature calculation unit 411 tracks the movement of the waste 50 in multiple pieces of temporally consecutive captured image information captured by the imaging unit 26, and calculates feature quantities that become input data for learning in the learning unit 412 described below. In other words, the feature calculation unit 411 divides the captured image information of the waste 50 input from the input unit 43 into multiple blocks, tracks at least one of the multiple divided blocks for a predetermined period of time, and calculates the feature quantities of the tracked block.

The feature calculation unit 411 calculates the temperature change amount of the waste 51 before it is supplied to the grate 4, for example, based on the captured image information obtained by photographing the waste 50 before it is supplied to the grate 4, i.e., the waste 51 before it is supplied located near the waste supply unit 12 (hereinafter also referred to as the "dust inlet area"). Specifically, this "temperature change amount" is the average temperature rise amount per block set near the waste supply unit 12 in the captured image information, as described below.

Here, FIG. 5 shows an example of photographed image information of the inside of the furnace 1 captured by the imaging unit 26. The photographed image information shown in the figure was divided into area A near the waste supply section 12 (dust inlet area), area B at the back of the grate 4, area C in the middle of the grate 4, area D at the front of the grate 4, area E on the left furnace wall 1a, and area F on the right furnace wall 1a, and the amount of change in the average temperature of each area A to F was calculated. The graph in FIG. 6 summarizes the progress of the average temperature of each area A to F, the progress of the total average temperature of each area A to F, and the progress of the calories measured after the waste before supply 51 related to the photographed image information is burned.

As shown in FIG. 6, among areas A to F, area A, i.e., area A near the waste supply section 12, has a relatively higher correlation between the average temperature and calories of the pre-supply waste 51 than the other areas. Based on this new knowledge, the feature calculation unit 411 calculates the amount of temperature rise of the pre-supply waste 51 near the waste supply section 12, which is highly correlated with the calories of the pre-supply waste 51, as a feature.

As described below, the feature calculation unit 411 calculates the feature amount of the waste 50 when the waste information prediction model 44a is learned by the learning unit 412 (see step S3 in FIG. 12) and when the waste information prediction unit 413 predicts the waste information (see step S13 in FIG. 13).

The specific method of calculating the features in the feature calculation unit 411 will be described below with reference to Figs. 7 to 9. As shown in Fig. 7, the feature calculation unit 411 divides the captured image information into a number of blocks Bl, and tracks at least one of the divided blocks Bl across a number of frames. As a method for tracking the movement of the waste 50, for example, optical flow using a block matching method can be used. In this method, the image is divided into blocks Bl of an appropriate size, and the difference between image frames is detected to search for the destination of each block Bl.

The blocks to be divided may be the same size for all of the captured image information, or only a portion of the image information may be divided into blocks of the same size. The block Bl division and tracking of the captured image information is intended to accurately calculate waste information (described below) without confusing a specific waste (such as the pre-combustion pre-supply waste 51 described above) with other waste (such as waste being burned), but if the block size is too small, processing will take a long time, and if it is too large, the tracking accuracy of the block Bl will decrease. For this reason, it is desirable to set the block size that provides the best processing time and tracking accuracy based on, for example, the resolution of the captured image information.

The feature amount calculation unit 411 places multiple blocks Bl in the area near the waste supply unit 12 in the captured image information, as shown in FIG. 7, for example. These blocks Bl move in accordance with the movement of the pre-supply waste 51, as shown in FIG. 8. In addition, any blocks that have fallen out of the tracking range set in the captured image information (the area near the waste supply unit 12 in this figure) are also deleted, and tracking is terminated. Furthermore, if the difference in pixel values of blocks Bl between frames is equal to or less than a predetermined threshold, it is determined that the pre-supply waste 51 contained in the block Bl has fallen into the grate 4, and tracking is terminated.

Next, the feature calculation unit 411 calculates the amount of change in average temperature over a specified time period for the tracked block Bl (tracked block) by analyzing the captured image information, as shown in FIG. 10. Next, the sum of the average temperatures of blocks Bl with a positive average temperature change (i.e., temperature is rising) is calculated. Note that a block Bl with a positive average temperature change means that the pre-supply waste 51 contained in that block Bl is likely to be combusted in the future.

Then, the feature calculation unit 411 calculates the average temperature rise per block as a feature by dividing the sum of the average temperatures of the blocks Bl with the positive average temperature change by the total number of blocks Bl. In this way, by looking at the temperature change of the pre-supply waste 51 tracked for each pre-set block Bl, rather than the entire area near the waste supply unit 12 in the captured image information, it is possible to accurately grasp the temperature change of the waste itself even if the pre-supply waste 51 moves.

Here, in addition to the temperature change amount of the pre-supply waste 51 for each block Bl described above, the feature calculation unit 411 may also calculate the movement amount of the pre-supply waste 51 for each block Bl as a feature amount. In this case, the feature calculation unit 411 calculates by tallying up statistics such as the average and variance of the motion vector of each block Bl. Note that, as shown in FIG. 11, when the movement amount of the block Bl is large, it means that the amount of waste 50 input into the furnace 1 is large. Conversely, when the movement amount of the block Bl is small, it means that the amount of waste 50 input into the furnace 1 is small. Therefore, by calculating the movement amount of the pre-supply waste 51 for each block Bl as a feature amount, it is possible to grasp the amount of waste 50 input into the furnace 1.

In addition to the temperature change amount of the pre-supply waste 51 for each block Bl and the movement amount of the pre-supply waste 51 for each block Bl described above, the feature calculation unit 411 may also calculate the total number of multiple blocks Bl into which the pre-supply waste 51 is divided and set in the captured image information as a feature amount. In addition, the number of fallen blocks that have fallen out of a tracking range (e.g., an area near the waste supply unit 12) set in advance in the captured image information represents the amount of garbage that will be burned, and therefore this may also be calculated as a feature amount.

The learning unit 412 uses the feature amounts calculated by the feature amount calculation unit 411, i.e., the feature amounts obtainable from each block Bl tracked by the feature amount calculation unit 411, as input variables and the waste information as output variables, and causes the waste information prediction model 44a to learn the relationship between past feature amounts and corresponding past waste information. The learning unit 412 then stores the waste information prediction model (learned waste information prediction model) 44a, which is a relational model that associates the feature amounts of pre-supply waste 51 with the waste information of the pre-supply waste 51, in the memory unit 44.

The learning method of the waste information prediction model 44a is not particularly limited, but machine learning methods suitable for learning time series data, such as RNN (Recurrent Neural Network), LSTM (Long Short Term Memory), and sequential linear prediction, can be used.

The waste information prediction model 44a is not limited to being a trained model. In other words, it may be an LUT (lookup table) or a specified function that shows the correspondence between the feature quantities of pre-supply waste 51 from the past incinerator that will predict and control the waste information from now on, with the feature quantities being input values, and the waste information of the pre-supply waste 51 being output values. When making a prediction, the feature quantities of the pre-supply waste 51 that are input may be approximated or interpolated appropriately using these LUTs or specified functions to calculate the waste information.

Here, the "feature quantities" learned by the waste information prediction model 44a include, as described above, at least one of the following: the amount of temperature change of the waste 50 for each block Bl, the amount of movement of the waste 50 for each block Bl, the total number of blocks Bl set in the captured image information, and the number of fallen blocks that have fallen out of a tracking range (e.g., the area near the waste supply section 12) previously set in the captured image information.

The "waste information" learned by the waste information prediction model 44a includes the quality of the waste 50. This "quality" also includes at least one of the calories (heat generation) generated when the waste 50 is burned and the amount of steam generated from the boiler 9 when the waste 50 is burned.

The waste information prediction unit 413 predicts waste information by inputting the feature values calculated by the feature value calculation unit 411 from the captured image information of the inside of the furnace 1 to the waste information prediction model 44a. For example, the waste information prediction unit 413 inputs the temperature rise amount of the pre-supply waste 51 for each block Bl near the waste supply unit 12 as input data to the waste information prediction model 44a, and obtains a predicted value of the quality (calories, water vapor amount) of the pre-supply waste 51 as an output.

The output unit 42 outputs predetermined information to the outside. In accordance with the control of the control unit 41, the output unit 42 displays an image of the waste 50 in the furnace 1, waste information (e.g., quality prediction results), etc. on the display monitor, displays characters, figures, etc. on the screen of the touch panel display, and outputs sound from the speaker. Specifically, the output unit 42 outputs waste information on the pre-supply waste 51 predicted by the waste information prediction unit 413.

The input unit 43 is configured using a user interface such as a keyboard, input buttons, levers, a touch panel for manual input superimposed on a display such as an LCD, and a microphone for voice recognition. A user operates the input unit 43 to input predetermined information to the control unit 41. The output unit 42 and the input unit 43 may be integrated into an input/output unit, and the input/output unit may be configured from a touch panel display, a speaker microphone, or the like. Specifically, the input unit 43 is configured to be able to input image information of the waste 50 photographed outside.

The memory unit 44 has a functional and physical configuration similar to that of the memory unit 32 described above, and is composed of storage media such as volatile memory such as RAM, non-volatile memory such as ROM, EPROM, HDD, and removable media.

The memory unit 44 can store an OS, various programs, various tables, various databases, and the like for executing the operation of the waste information prediction device 40. Here, the various programs include an information processing program that realizes control using the learning model or the learned model according to this embodiment. The memory unit 44 may be provided in another server that can communicate via various networks, or may be provided in the combustion control device 30. Specifically, the memory unit 44 stores a waste information prediction model 44a.

The waste information prediction model 44a is a model trained by the learning unit 412, and is a model that predicts waste information (e.g., the quality of the waste 51 before supply) from features (e.g., the amount of temperature rise of the waste 51 before supply) acquired from each block Bl of the photographed image information. The waste information prediction model 44a is trained to output waste information when features calculated from a tracking block that tracks multiple blocks Bl set in the photographed image information of the waste 51 before supply are input.

The waste information prediction model 44a is intended to be used, for example, as a program module that is part of artificial intelligence software, and is used in a computer (waste information prediction device 40) that has a CPU and a storage device. The waste information prediction model 44a may also be recorded on a recording medium such as a CD, DVD, flash memory, or magnetic tape, and made readable.

(Learning method for waste information prediction model)
A learning method for a waste information prediction model according to an embodiment of the present invention will be described with reference to Fig. 12. The learning method for a waste information prediction model includes an image acquisition step, a feature amount calculation step, and a learning step.

First, in the image acquisition process, the imaging unit 26 captures an image of the waste 50 in the incinerator furnace 1 and acquires captured image information (step S1). Next, in the feature amount calculation process, the feature amount calculation unit 411 divides the captured image information acquired in step S1 into multiple blocks Bl (step S2). Next, the feature amount calculation unit 411 tracks the movement of the waste 50 for each block Bl and calculates a feature amount (e.g., the temperature rise amount of the waste 51 before supply) that characterizes the waste 50 contained in each block Bl (step S3). Next, in the learning process, the learning unit 412 trains the waste information prediction model 44a to learn the relationship between the feature amount calculated in step S3 and the waste information (e.g., the quality of the waste 51 before supply) (step S4), and this flow is completed.

(Waste information forecast method)
A waste information prediction method according to an embodiment of the present invention will be described with reference to Fig. 13. The waste information prediction method includes an image acquisition step, a feature amount calculation step, and a waste information prediction step.

First, in the image acquisition process, the imaging unit 26 captures an image of the waste 50 in the incinerator furnace 1 and acquires captured image information (step S11). Next, in the feature amount calculation process, the feature amount calculation unit 411 divides the captured image information acquired in step S11 into a plurality of blocks Bl (step S12). Next, the feature amount calculation unit 411 tracks the movement of the waste 50 for each block Bl and calculates a feature amount (e.g., the temperature rise amount of the waste 51 before supply) that characterizes the waste 50 contained in each block Bl (step S13). Next, in the waste information prediction process, the waste information prediction unit 413 inputs the feature amount calculated in step S13 into the waste information prediction model 44a to predict waste information (e.g., the quality of the waste 51 before supply) (step S14), and this flow is completed.

According to the waste information prediction device, incinerator combustion control device, waste information prediction method, waste information prediction model learning method, and waste information prediction model program of the embodiments described above, the movement of waste 50 in the furnace 1 is tracked for each of a number of pre-set blocks Bl, and the waste information prediction model 44a is trained on the features of each block Bl, thereby making it possible to accurately predict waste information in the furnace 1, and to efficiently control waste treatment plants such as incinerators based on the predicted waste information.

In other words, according to the waste information prediction device, incinerator combustion control device, waste information prediction method, waste information prediction model learning method, and waste information prediction model program of the embodiments, the movement of waste 50 is tracked in detail from captured image information, and the change over time in the temperature rise of the waste 50 of interest is captured, thereby extracting feature quantities that are highly correlated with the quality of the waste 50 (e.g. calories, water vapor content). Learning and prediction are then performed using the feature quantities as input data, and the prediction results are used to control the waste treatment plant, thereby making it possible to achieve high accuracy and efficiency in the control of the waste treatment plant.

For example, if the amount of temperature rise of the pre-supply waste 51 is small (the rate of temperature rise is slow), it is estimated that the pre-supply waste 51 contains a lot of moisture and has low calories. On the other hand, if the amount of temperature rise of the pre-supply waste 51 is large (the rate of temperature rise is fast), it is estimated that the pre-supply waste 51 contains little moisture and has high calories. In this way, in the waste information prediction device, waste information prediction method, and waste information prediction model learning method according to the embodiment, the relationship between the amount of temperature rise and quality of the pre-supply waste 51 can be learned in advance by the waste information prediction model 44a, thereby making it possible to predict the quality of the pre-supply waste 51 with high accuracy.

In addition, in the incinerator combustion control device and incinerator combustion control method according to the embodiment, for example, when the predicted calorie is high, control is performed to reduce the amount of waste 50 placed on the grate 4 (i.e., the waste supply device feed speed), and when the predicted calorie is low, control is performed to increase the amount of waste 50 placed on the grate 4. This makes it possible to stabilize the combustion of the waste 50 and realize stabilization of the amount of steam and power generation generated by the boiler 9.

Further advantages and modifications may be readily derived by those skilled in the art. The broader aspects of the present disclosure are not limited to the specific details and representative embodiments shown and described above. Thus, various modifications may be made without departing from the spirit or scope of the general inventive concept as defined by the appended claims and equivalents thereof.

REFERENCE SIGNS LIST 1 furnace 1a furnace wall 2 waste inlet 3 waste supply device 4 grate 5 ash drop port 6 combustion air blower 7 furnace outlet 8 chimney 9 boiler 9a heat exchanger 9b steam drum 10 secondary air inlet 11 secondary air blower 12 waste supply section 13 step wall 14 combustion air damper 14a, 14b, 14c, 14d under-grate combustion air damper 15 secondary air damper 16 intermediate ceiling 17 combustion chamber gas thermometer 18 main flue gas thermometer 19 furnace outlet lower gas thermometer 20 furnace outlet middle gas thermometer 21 furnace outlet gas thermometer 22 boiler outlet oxygen concentration meter 23 gas concentration meter 24 exhaust gas flow meter 25 steam flow meter 26 imaging section 30 combustion control device 31 control section 32 Memory unit 33 Operation amount adjustment unit 331 Combustion air adjustment unit 332 Air amount ratio adjustment unit 333 Secondary air adjustment unit 334 Waste supply device feed rate adjustment unit 335 Grate feed rate adjustment unit 40 Waste information prediction device 41 Control unit 411 Feature amount calculation unit 412 Learning unit 413 Waste information prediction unit 42 Output unit 43 Input unit 44 Memory unit 44a Waste information prediction model 50 Waste 51 Waste before supply 52 Waste on grate

Claims (17)

A waste information prediction device that predicts information about waste in an incinerator,
An input unit capable of inputting photographed image information of the waste photographed outside;
a feature amount calculation means for dividing the photographed image information of the waste material input from the input unit into a plurality of blocks, tracking at least one of the divided blocks for a predetermined period of time, and calculating a feature amount of the tracking block;
a storage unit that stores a relational model that associates the feature quantities of the waste with information about the waste;
A control means;
having
The control means calculates features of the waste from image information obtained by photographing the waste inside the incinerator and inputted from the input unit using the feature calculation means, and predicts information about the waste based on the relational model stored in the memory unit. This is a waste information prediction device characterized by the above. The waste information prediction device according to claim 1, characterized in that the feature amount includes at least one of the amount of temperature change of the waste, the amount of movement of the waste, and the total number of multiple blocks into which the waste is divided and set in the captured image information. The waste information prediction device according to claim 1 or 2, characterized in that the information about the waste includes at least one of the calories generated when the waste is burned and the amount of steam generated from the boiler of the incinerator when the waste is burned. The waste information prediction device according to any one of claims 1 to 3, characterized in that it has an output unit that outputs information about the waste predicted by the control means. The waste information prediction device according to any one of claims 1 to 4, characterized in that the waste image information input from the input unit is image information obtained by photographing the waste before it is supplied to the grate of the incinerator. The waste information prediction device according to any one of claims 1 to 5, characterized in that the relationship model stored in the storage unit is a learned waste information prediction model obtained by learning from the relationship between past waste features and information about the past waste. The waste information prediction device according to any one of claims 1 to 5, characterized in that the relational model stored in the storage unit is a lookup table or a predetermined function in which the feature quantities of past waste are input values and information about the past waste is output values. A combustion control device including a combustion control unit that controls an incinerator that burns waste,
The combustion control unit is
A combustion control device for an incinerator, comprising: a control device for controlling combustion in the incinerator based on information about waste output from the waste information prediction device according to any one of claims 4 to 7. The incinerator includes a grate for moving the waste, a waste supply device for supplying the waste onto the grate, and a blower for blowing air into the incinerator;
The combustion control device for an incinerator as described in claim 8, characterized in that the combustion control unit controls at least one of the following based on information about the waste: the movement speed of the waste on the grate, the supply speed of the waste by the waste supply device, the amount of air blown by the blower, and the temperature of the air by the blower. A waste information prediction method for predicting information about waste in an incinerator, comprising:
An image acquisition step of acquiring photographed image information of waste put into the incinerator;
a feature amount calculation step of dividing the photographed image information of the waste obtained in the image acquisition step into a plurality of blocks, tracking at least one of the divided plurality of blocks, and calculating a feature amount of the tracking block;
a waste information prediction step of predicting information about the waste from the feature amounts of the waste calculated in the feature amount calculation step;
A waste information prediction method comprising the steps of: The waste information prediction method according to claim 10, characterized in that the waste information prediction step predicts using a relational model in which the waste characteristics and information about the waste are previously associated with each other. The waste information prediction method according to claim 11, characterized in that the relationship model is a learned waste information prediction model obtained by learning from the relationship between past waste features and information about the past waste. The waste information prediction method according to claim 11, characterized in that the relational model is a lookup table or a predetermined function in which the feature quantities of past waste are input values and information about the past waste is output values. The waste information prediction method according to any one of claims 10 to 13, characterized in that the photographed image information of the waste acquired in the image acquisition process is photographed image information obtained by photographing the waste before it is supplied to the grate of the incinerator. The waste information prediction method according to any one of claims 10 to 14, further comprising a combustion control step of controlling combustion in the incinerator based on the waste information predicted in the waste information prediction step. A method for learning a waste information prediction model that predicts information about waste in an incinerator, comprising:
Dividing photographed image information of waste in an incinerator for incinerating the waste into a plurality of blocks, and tracking at least one of the plurality of divided blocks;
A learning method for a waste information prediction model, characterized in that the feature values calculated from the tracking block are used as input values, information about the waste is used as output values, and the waste information prediction model is made to learn the relationship between the past feature values and the corresponding past information about the waste. A waste information prediction model program that can be recorded on a recording medium and predicts information about waste in an incinerator,
A waste information prediction model program that is trained to output information about the waste when a feature calculated from a tracking block that tracks multiple blocks set in the captured image information of the waste is input.

Images