JPH0926819A - Plant abnormality diagnosis device - Google Patents

Abstract

(57) [Summary] (Correction) [Problem] To provide a plant abnormality diagnosing apparatus for rapidly detecting the occurrence of a plant abnormality from two variables having no first-order correlation. A plant abnormality diagnosis apparatus includes a process data collection unit (11) that periodically collects data, a differential calculation unit (12) that differentiates a process state quantity and an operation quantity within a minute time twice or once with respect to time, respectively. Differentiating unit 12
The operation result storage unit 13 for accumulating each data obtained in the above as a time-series numerical data pair relating to the process state quantity and the manipulated variable, and taking out a predetermined number of numerical data pairs for statistical processing,
An area definition unit 14 that defines a data area corresponding to a data pair in a normal operating state, a data pair obtained in real time from the calculation result storage unit 13 and a predefined data area are used to determine whether the control system is normal or not. An abnormality determination unit 15 that determines an abnormality, and an abnormality cause estimation unit 16 that estimates an abnormality cause from the deviation direction of a data pair from the data area are provided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a plant abnormality diagnosing apparatus for promptly detecting a symptom of abnormality in a plant and giving information to a plant operator.

[0002]

2. Description of the Related Art In recent years, the operation of chemical plants has become more sophisticated and the number of plant operators has been reduced for the purpose of rationalization. Therefore, various alarm functions are added to the distributed instrumentation system (DCS) for the purpose of supporting plant operators. For example, temperature, pressure,
Examples include upper / lower limit alarms, rate-of-change alarms and the like regarding process state quantities such as liquid level and flow rate.

As an example of the above-mentioned alarm, the correlation value of two process signals consisting of a process state quantity and an operation quantity for controlling it is constantly monitored, and when the correlation value exceeds a certain threshold value, it is regarded as abnormal. , There is a technology to alert plant operators. JP-A-4-186119 and JP-A-3-5
Japanese Patent No. 3123 describes an example of a technique for monitoring such correlation. Here, Japanese Patent Laid-Open No. 4-186119
Japanese Unexamined Patent Application Publication No. Hei 3 (1999) describes a monitoring technique for early detection of a process abnormality by monitoring a deviation between a reference value obtained from the correlation between process signals to be monitored and a process signal. According to Japanese Patent Laid-Open No. 53123/1993, when the data captured online exceeds a preset threshold value, it is determined from the past time series data whether or not there is an abnormality in the way the threshold value is exceeded. Techniques are described that reduce the number of alarms and increase the reliability of monitoring.

[0004]

Among the above-mentioned conventional techniques, the technique of monitoring the upper limit / lower limit of the process state quantity is
There is a problem that changes in operating conditions cannot be handled.
Further, the technique of monitoring the correlation value can be adopted when there is a first-order correlation between the process state quantity and the manipulated value thereof like a flow rate control system, but, for example, as in the case of a liquid level control system. However, there is a drawback that it cannot be adopted if there is no first-order correlation between the process state quantity and the manipulated variable.

In view of the above, it is an object of the present invention to provide a plant abnormality diagnosing apparatus capable of detecting a process abnormality in a control system at an early stage in which there is no first-order correlation between a process state quantity and an operation amount thereof. And

[0006]

In order to achieve the above object, the present invention collects each data in a minute time of a process state quantity of a control system to be monitored and an operation amount for controlling the process state quantity. Process data collecting section for periodically collecting data, and differential calculating section for differentiating the collected data of the process state quantity and the manipulated quantity within a minute time twice or once respectively with respect to time and outputting them as numerical data. , A calculation result storage unit for accumulating numerical data obtained by a differential calculation unit as a time-series data pair relating to the process state quantity and the operation amount, a predetermined number of data pairs are taken out and statistically processed, and data of a normal operating state An area definition unit that defines a data area corresponding to the pair, and a data pair obtained in real time from the calculation result storage unit, and based on the previously defined data area Provides a plant abnormality diagnostic device, characterized in that it comprises an abnormality determining section to determine normal or abnormal in the control system.

In the plant abnormality diagnosing apparatus of the present invention, the process data collecting unit first collects data of the process state quantity of the control system to be monitored and the operation amount for controlling the process state quantity within a minute time, The collected process state quantity and the corresponding manipulated variable are each differentiated twice or once with respect to time by the differential calculation unit to obtain numerical data. This data collection and differentiation are periodically performed, and each obtained numerical value data is stored in the calculation result storage unit as a time series data pair regarding the process state quantity and the operation quantity. The area definition unit extracts a predetermined number of data pairs from this, performs statistical processing, and predefines a data area corresponding to the data pair in a normal operating state prior to monitoring. The abnormality determination unit retrieves the data pair from the operation result storage unit in real time, compares the data pair with a predefined data area, and when the data pair deviates from the data area for a predetermined number of times in succession, for example. , It is determined that there is an abnormality in the control system.

With the above operation, even in a control system in which there is no primary correlation between the process state quantity and the manipulated value, an abnormal state is automatically generated during plant operation without burdening the operator. Can be detected. Particularly, the plant abnormality diagnosing device of the present invention has a 90
It is applied to control systems with a phase difference of degrees.

In the plant abnormality diagnosing device of the present invention, it is preferable that an abnormality cause estimating unit is further provided, and the abnormality cause estimating unit refers to, for example, a knowledge database regarding the control system determined to be abnormal by the abnormality determining unit. , Estimate the cause of abnormality based on the deviation direction of the data pair from the data area. In addition to this, it is also preferable to output a possible coping method for the cause of the abnormality.

[0010]

BEST MODE FOR CARRYING OUT THE INVENTION The present invention will be described in more detail below with reference to the drawings based on an embodiment in which the present invention is applied to a liquid level control system. FIG. 1 is a block diagram of a plant abnormality diagnosis apparatus according to an embodiment of the present invention.

A liquid level gauge is installed in the liquid level control system 1 to be monitored for abnormalities, and the liquid level gauge detects the liquid level of the tank in the plant. The detection signal of the liquid level gauge that constitutes the process state quantity is distributed type instrumentation system (DCS).
2 to the computer 10 and the process data collecting unit 11 in the computer 10. Further, the opening degree of the control valve for supplying the liquid to this tank is also given from the liquid level control system 1 to the process data collecting section 11 as an operation amount via the distributed instrumentation system 2. The process collection unit 11 collects a plurality of data of process state quantities and operation quantities within a minute time. This data collection is further periodically performed, and the obtained data is given to the differential operation section 12 each time.

The collected process state quantity within a minute time and the corresponding manipulated variable are differentiated twice or once with respect to time by the differential operation unit 12, respectively, and numerical values relating to the process state quantity and manipulated quantity within a minute time. It is output as data. This numerical data is periodically output and accumulated in the calculation result storage unit 13 as time-series numerical data.

Generally, the manipulated variable (opening of the control valve) in the liquid level control system is generally the phase of the process state quantity (liquid level) because the liquid level control system is an integral process.
The phase is shifted by 90 °, and the two cannot be simply correlated. Therefore, in the present invention, it is considered that the process state quantity is subjected to time differentiation to remove the phase difference and obtain the correlation between the two. Furthermore, the rate of change in the manipulated variable eliminates the variation in the distribution of the manipulated variable that accompanies changes in the operating rate
Focusing on facilitating abnormality monitoring from the scatter diagram, for each time derivative of process state quantity and manipulated variable,
Further, time differentiation was performed to convert both data into numerical data, and then the data were correlated. In the present invention, after all, the abnormality data is monitored by obtaining the correlation between the numerical data of the two-time differential value of the process state quantity and the one-time differential value of the manipulated variable.

In accordance with the above, the differential operation section 12 differentiates the liquid level and the opening of the control valve twice with respect to time and once, respectively, and stores the operation results as a time-series numerical data pair. Give to part 13. Area definition part 1
Reference numeral 4 designates a range of data pairs corresponding to a normal operating state as a data area by taking out a predetermined number of the time-series data pairs from the operation result storage unit 13 during the operation of the plant and statistically processing these. To do.

FIG. 2 is a graph illustrating the data area (monitoring area) 20 obtained as described above. The data pair consisting of the two-time differential value of the liquid level and the one-time differential value of the control valve opening is represented as a point on the scatter diagram, and a predetermined number of data pairs are shown on the graph. As shown in the figure, the data pairs have a relatively strong correlation with each other. The data pair is statistically processed, and the data area 20 which is the range of the data pair in the normal operation state is obtained as the Mahalanobis general distance circle of the data pair. Here, as the actual calculation,
First, the center of gravity and the standard deviation of the normal data pair are obtained, and then the data area 20 in which the distance from the center of gravity is determined based on the standard deviation is obtained. This data area 2
0 is used for determining whether the control system is normal / abnormal in the subsequent operation, and can be displayed on the CRT display 3 in FIG.

After the data area 20 is defined, it is determined whether or not there is an abnormality in the control system. FIG. 3 is a flowchart showing this procedure. Step S1 is a step in which a data area is defined by the data area definition unit. Then, from the liquid level control system 1 to be monitored, via the distributed instrumentation system 2, the process data collection unit 11
Data including the process state quantity and the manipulated variable is collected in the same manner as described above (step S2).

Similarly, the above-described twice differentiation (process state quantity) and once differentiation (manipulation quantity) are performed on the obtained process data in each minute time in the differential operation section 12 (step S3). ). The abnormality determination unit 15 determines whether each data pair obtained by the differential operation unit 12 is in the previously defined data area 20 or deviates from the data area 20. The abnormality determination unit 15 determines that there is an abnormality when the deviation of the data pair from the data area exceeds a predetermined number of times in succession (step S4).

As described above, the determination of normality / abnormality is made based on the presence / absence of continuous deviation of the data pair from the data area 20 a preset number of times. By monitoring the continuous deviations, it is possible to prevent false alarms due to noise included in the data and to generate alarms only when the control system is abnormal. As a result of the determination by the abnormality determination unit 15, an alarm message is displayed or printed on the CRT display 3 or the printer 4 and the buzzer 5 sounds for the liquid level control system which is diagnosed as having an abnormality. As a result, the plant operator is notified of the presence of an abnormality in the liquid level control system (step S5).

Subsequently, the abnormality cause estimating unit 16 detects the deviation direction of the data pair from the data area and refers to the knowledge database 17 to estimate the cause of the abnormality occurrence in the control system (step S6). The estimated cause of abnormality is output by the CRT display 3 and the printer 4 together with the countermeasure for the abnormality contained in the knowledge database 17 (step S7). The knowledge database 17 stores knowledge data in which a deviation direction of a data pair from a data area, a candidate of an abnormality cause, and a candidate of a countermeasure for the abnormality are described in association with each other.

The cause of abnormality is estimated by the knowledge database 17 as follows. 4 and 5 are graphs for explaining the state of this estimation, FIG. 4 exemplifies the deviation direction from the data area of the data pair, and FIG. 5 is the abnormality cause estimated corresponding to each deviation direction. Is shown. It is now assumed that a predetermined number of rightward continuous deviations from the data area are confirmed on the scatter diagram as shown in FIG. In the abnormality diagnosing device, the abnormality cause estimating unit 16 for estimating the cause of the abnormality is activated after issuing the alarm of the abnormality occurrence. Here, in the knowledge database 17, as shown in FIG. 5, the deviation directions on the scatter diagram are (1) for the right direction, (2) for the left direction, (3) for the up direction, and () for the down direction. 4), and the numbers in that direction, the causes of abnormalities, and the candidates for their countermeasures are described in association with each other as shown in Table 1.

[0021]

[Table 1]

Here, the continuous deviation of the data pair in the right direction (1) is presumed to be "absence of detection signal due to liquid level change" as the cause of abnormality according to Table 1, and that fact is indicated. Displayed at 4. In addition, as a basis for this, "the displacement is not picked up in the transmitter, so there is no change in the deviation, and the valve opening gradually increases. Therefore, compared to the steady state, the valve changes with respect to the changing speed of the liquid level. There is a display saying `` The opening tends to be large '', and as a countermeasure, for example, if the type of detector is a displacer type, check "whether float, knife edge sticking" etc. It is also displayed. The causes of deviations in other directions are similarly estimated. These are displayed on the CRT display 3 and also printed on the printer 4.

As described above, in the abnormality diagnosing device of this embodiment, not only the occurrence of the plant abnormality but also the cause of the abnormality and the countermeasure are displayed on the CRT display 3 and the printer 4. As a result, the abnormal state of the plant can be quickly detected before the failure, and the plant operator can take prompt action against the abnormal state.

In the above embodiment, when the data area is set, by making the data area large, that is, by taking a large Mahalanobis general distance from the data center, it is possible to prevent the occurrence of false alarm. Therefore, for the data area, for example, a numerical value such as three times the standard deviation is adopted as the distance from the center of gravity. When there is an abnormality, the degree of departure from the data area is large compared to the normal operating state, so it is possible to detect an abnormality in the control system even if a somewhat large data area is defined.

Although the present invention has been described above based on its preferred embodiments, the present invention can be modified and changed in various ways from the above-described embodiments, and various modifications and changes have been made from the above-described embodiments. A plant abnormality diagnosis device is also included in the scope of the present invention.

[0026]

As described above, according to the plant abnormality diagnosing apparatus of the present invention, a control system such as a liquid level control system having a variable that cannot be correlated between the process state quantity and the manipulated value thereof. In this case, the correlation can be monitored, the occurrence of the process abnormality can be detected early, and the burden on the operator can be reduced. Further, in addition to this, by informing the operator of the estimation of the cause of the abnormality and the coping method thereof, it is possible to quickly perform the processing for the abnormality occurrence. Since the data area can be set automatically from a predetermined number of data pairs during normal operation, no burden is imposed on the operator.

[Brief description of drawings]

FIG. 1 is a block diagram of a plant abnormality diagnosis apparatus according to an embodiment of the present invention.

FIG. 2 is a graph showing how data areas are set by the abnormality diagnosis device of FIG.

FIG. 3 is a flowchart showing processing in the abnormality diagnosis device of FIG.

FIG. 4 is a graph illustrating the deviation of a data pair from the data area.

FIG. 5 is an explanatory diagram for estimating a cause of abnormality based on a deviation direction of a data pair from a data area.

[Explanation of symbols]

 1 Liquid Level Control System 2 Distributed Instrumentation System 3 CRT Display 4 Printer 5 Buzzer 10 Computer 11 Process Data Collection Section 12 Differential Calculation Section 13 Calculation Result Storage Section 14 Area Definition Section 15 Abnormality Determination Section 16 Abnormality Cause Estimation Section 17 Knowledge Database 20 data areas

Claims (2)

[Claims] 1. A process data collecting unit that periodically collects data for collecting each data of a process state quantity of a control system to be monitored and an operation amount for controlling the process state quantity within a minute time, A differential operation unit that differentially or once differentiates each data of the process state amount and the operation amount in a minute time with respect to time, and outputs as numerical data. The numerical data obtained by the differential operation unit is used as the process state amount and Calculation result storage unit that accumulates as time-series data pairs related to manipulated variables, area definition unit that extracts a predetermined number of data pairs and performs statistical processing, and defines a data area corresponding to a data pair in a normal operating state, and calculation results An abnormality determination unit that determines whether the control system is normal or abnormal based on a data pair obtained in real time from the storage unit and the previously defined data area A plant abnormality diagnosis device characterized by the following. 2. The control system further includes an abnormality cause estimation unit that estimates an abnormality cause of the control system determined to be abnormal by the abnormality determination unit based on a deviation direction of a data pair from the data area.
The plant abnormality diagnosis device according to claim 1.