JP7232028B2 - Operation monitoring device and method - Google Patents

Description

The present invention relates to an operation monitoring technique for monitoring the operating status of a batch process.

Generally, in a plant system, a so-called batch process is executed, in which the same process is repeatedly executed for each manufacturing unit, with a fixed amount of raw materials or products being manufactured. In a batch process, various plant equipment is automatically operated based on a preset batch sequence. In a batch process, if an abnormality occurs in any plant equipment, the abnormality may affect the entire plant system. For this reason, in plant systems, an operation monitoring device is used to monitor the operating status of a batch process based on various process variables obtained from the batch process.

Conventionally, as such an operation monitoring device, a technique for monitoring the operation status of a batch process based on the operation time of plant equipment has been proposed (see Patent Document 1, etc.). Based on the operation signal acquired from the plant equipment, the operation time from the start to the end of the operation is collected for each batch operation, the reference range for the operation time is derived in advance, and the batch operation to be monitored is calculated. If the operation time obtained from the above deviates from the reference range, it is diagnosed that there is a sign of abnormality in the plant equipment.

Japanese Patent Application Laid-Open No. 2003-131729

The above-mentioned prior art is based on the premise that an abnormality sign of a batch process appears in advance in the change in operation time from the start of operation to the end of operation of plant equipment, and the operation time at the time of normal completion obtained in the past is used as a judgment criterion. It is used. However, the method of detecting an anomaly sign based on such a premise cannot detect an anomaly sign such as a change that differs from the normal state in the process quantity that indicates the actual operating status of the batch process. was there.

For example, in the case of a batch process that uses chemical reactions, the reaction temperature and reaction pressure can suddenly change due to factors such as raw material charging, heat removal, agitation, and operation. Accidents have been reported. In such a case, there is a significant anomaly sign that the process variables, such as the reaction temperature and the reaction pressure, which indicate the actual operating conditions of the batch process deviate greatly from their target values. In such cases, it is difficult to accurately detect signs of abnormality based on the operation time from the start of operation of plant equipment to the end of operation.

Moreover, in order to detect an anomaly sign based on the operation time of plant equipment, it is necessary to wait until at least the normal completion time elapses after the plant equipment starts operating. Therefore, it is not only impossible to detect an anomaly sign of a batch process at an early stage, but there is also the possibility that the detection of an anomaly sign will be delayed, resulting in a serious situation.

Moreover, in general, the quantity and quality of raw materials, cooling water temperature, air temperature, and other operating conditions are not constant in individual batch processes. However, when such operating conditions change, the operating time of the plant equipment greatly fluctuates. Therefore, if the abnormal sign of the batch process that appears in the operating time is smaller than the fluctuation range of the operating time that fluctuates due to these various factors, the detection of the abnormal sign based on the operating time of the plant equipment will not detect such a case. Accurate detection is difficult.

The present invention is intended to solve such problems, and an object of the present invention is to provide an operation monitoring technique capable of accurately and early detection of signs of abnormalities in a batch process.

In order to achieve such an object, an operation monitoring apparatus according to the present invention monitors the operation status of a batch process at each monitoring timing based on the process amount to be monitored obtained from the batch process. a target value trajectory calculation unit that calculates a target value trajectory indicating the target value at the monitoring timing based on a transition of the target value related to the process variable set in advance; a process variable prediction unit that predicts a future predicted value of the process variable at a future monitoring timing included in a future interval based on time-series data; and based on the future predicted value and the target value at the future monitoring timing. and an anomaly sign detection unit that detects an anomaly sign that may occur in the future interval.

Further, one configuration example of the operation monitoring device according to the present invention further includes a change speed calculation unit that calculates a change speed indicating a change amount per unit time of the future prediction value at the future monitoring timing, The anomaly sign detection unit compares the rate of change with a preset threshold value at each of the future monitoring timings, and detects a sign of an anomaly that may occur in the future interval based on the obtained comparison result. It is designed to detect.

Further, another operation monitoring device according to the present invention is an operation monitoring device that monitors the operation status of the batch process at each monitoring timing based on the process amount to be monitored acquired from the batch process. a target value trajectory calculating unit that calculates a target value trajectory indicating the target value at the monitoring timing based on the transition of the target value related to the process variable obtained in the past; , a process variable prediction unit that predicts a future predicted value of the process variable at a future monitoring timing included in the future interval; A change speed calculator that calculates the change speed and a preset threshold value are compared for each of the future monitoring timings, and an abnormality that can occur in the future interval is determined based on the obtained comparison result. and an anomaly sign detection unit that detects a sign.

Further, one configuration example of the operation monitoring apparatus according to the present invention further includes an alarm output unit that outputs an alarm indicating that a sign of an abnormality that may occur in the future section has been detected, and the abnormality sign detection unit and outputting an alarm from the alarm output unit when the deviation between the future predicted value and the target value at the future monitoring timing is larger than a predetermined allowable deviation.

Further, one configuration example of the operation monitoring apparatus according to the present invention further includes an alarm output unit that outputs an alarm indicating that a sign of an abnormality that may occur in the future section has been detected, and the abnormality sign detection unit and, when the speed of change exceeds the threshold value, the alarm output section is caused to output an alarm.

Further, in one configuration example of the operation monitoring device according to the present invention, the process variable prediction unit predicts the past monitoring from the time-series data of the process variable obtained in the past and the transition of the target value related to the process variable. Based on the deviation between the process variable and the target value obtained at each timing, the deviation at the future monitoring timing is predicted for each of the future monitoring timings, and the deviation is calculated from the target value at the future monitoring timings. By subtracting, the future prediction value at the future monitoring timing is calculated.

Further, the operation monitoring method according to the present invention includes a target value trajectory calculation unit, a process amount prediction unit, and an abnormality sign detection unit. An operation monitoring method used in an operation monitoring device for monitoring the operation status of a batch process, wherein the target value trajectory calculation unit determines the target value at the monitoring timing based on the transition of the target value related to the process variable set in advance a target value trajectory calculation step of calculating a target value trajectory indicating a value; a process variable prediction step of predicting a future predicted value of the process variable at the future monitoring timing; and an anomaly sign detection step of detecting an anomaly sign that may occur in the future interval based on the target value.

Another operation monitoring method according to the present invention includes a target value trajectory calculation unit, a process variable prediction unit, a change speed calculation unit, and an abnormality sign detection unit, and is based on a process variable to be monitored obtained from a batch process. an operation monitoring method for use in an operation monitoring device for monitoring the operation status of the batch process at each monitoring timing, wherein the target value trajectory calculation unit calculates the desired value based on the transition of the preset target value related to the process variable a target value trajectory calculating step of calculating a target value trajectory indicating the target value at the monitoring timing; a process variable prediction step of predicting a future predicted value of the process variable at the future monitoring timing for each future monitoring timing; a change rate calculation step of calculating a rate of change indicating an amount of change per unit time of the predicted future value in the above; and an anomaly sign detection step of comparing with a threshold value and detecting an anomaly sign that may occur in the future interval based on the obtained comparison result.

According to the present invention, when the future predicted value of the process variable deviates from the target value of the target value trajectory by more than the allowable deviation, or when the change speed of the future predicted value deviates from the allowable range, it is immediately regarded as an anomaly sign. It will be detected and an alarm will be output. Therefore, for each monitoring timing, it is possible to confirm the presence or absence of an abnormality that may occur in the future interval from the monitoring timing, and it is possible to accurately and early detect a sign of abnormality in the batch process.

FIG. 1 is a block diagram showing the configuration of the operation monitoring device. FIG. 2 is an explanatory diagram showing an example of generation of target value trajectory data. FIG. 3 is an explanatory diagram showing a calculation example of the future prediction value using the target process quantity. FIG. 4 is an explanatory diagram showing a monitoring screen display example corresponding to the calculation example of FIG. FIG. 5 is an explanatory diagram showing a calculation example of a future prediction value using deviation. FIG. 6 is an explanatory diagram showing a monitoring screen display example corresponding to the calculation example of FIG. FIG. 7 is an explanatory diagram showing an example of conversion speed calculation. FIG. 8 is a flowchart showing operation monitoring processing. FIG. 9 is an explanatory diagram showing an example of a monitor screen display showing changes in deviation. FIG. 10 is an explanatory diagram showing an example of a monitor screen display showing changes in future prediction values.

Next, one embodiment of the present invention will be described with reference to the drawings.
[Operation monitoring device]
First, a driving monitoring device 10 according to the present embodiment will be described with reference to FIG. FIG. 1 is a block diagram showing the configuration of the operation monitoring device.
The operation monitoring device 10 is composed of an information processing device such as a PC and a server device as a whole, and is a device that monitors the operating status of a batch process based on the process amount obtained from the process equipment and detects signs of abnormality.

The controller 21 is composed of a control device such as an industrial controller as a whole, controls field devices such as a sensor 22 and an actuator 23 connected under it via a communication line L2, and detects process variables detected by these field devices. to the host device 20 and the operation monitoring device 10 via the communication line L1.
The host device 20 is composed of an information processing device such as a server device as a whole, manages the operation of the controller 21 via the communication line L1, and transmits the process quantity acquired from the controller 21 to the operation monitoring device 10 via the communication line L1. It has a function to deliver to

As shown in FIG. 1, the operation monitoring device 10 includes a communication I/F section 11, an operation input section 12, a screen display section 13, a storage section 14, and an arithmetic processing section 15 as main components.

The communication I/F unit 11 has a function of performing data communication with external devices such as the host device 20 and the controller 21 via the communication line L1.
The operation input unit 12 includes an operation input device such as a keyboard, a mouse, and a touch key, and is a device that detects input operator operations and outputs them to the arithmetic processing unit 15 .
The screen display unit 13 is a screen display device such as an LCD, and is a device that displays various screens such as a menu screen, a setting screen, an input screen, and an output screen output from the arithmetic processing unit 15 .

The storage unit 14 is a storage device such as a hard disk or a semiconductor memory, and is a device that stores various processing data and programs 14P used for operation monitoring processing executed by the arithmetic processing unit 15 .
The program 14P is a program that executes operation monitoring processing by cooperating with the CPU of the arithmetic processing unit 15, and is read in advance from an external device or a storage medium connected to the operation monitoring device 10 and stored in the storage unit 14. Stored.

Main processing data stored in the storage unit 14 include operation data 14A, target value trajectory data 14B, process amount data 14C, and predicted value data 14D.

The operation data 14A is data indicating the operation details of the batch process, and includes changes in the target value of each target process quantity to be monitored.

The target value trajectory data 14B is time-series data indicating changes in the target value of the target process variable at preset monitoring timings. In the following, a case will be described as an example in which a timing that arrives at a constant monitoring cycle is used as the monitoring timing, but the present invention is not limited to this. may

The process amount data 14C is data indicating the target process amount, which is the process amount to be monitored in the batch process, acquired from the host device 20 or the controller 21 . The process amount data 14C sequentially stores the target process amount acquired from the batch process in accordance with the start of the batch process.

The predicted value data 14D is time-series data indicating predicted values in the future interval from the time when the new target process amount is acquired to the end of the batch process. The predicted value data 14D includes a future predicted value of the target process amount predicted at each monitoring timing and a change rate predicted value indicating the amount of change per unit time of the future predicted value.

The arithmetic processing unit 15 is a circuit unit that is composed of a CPU and its peripheral circuits, and realizes various processing units that perform operation monitoring processing through cooperation between the CPU and the program 14P.
Main processing units implemented by the arithmetic processing unit 15 include a target value trajectory calculation unit 15A, a data acquisition unit 15B, a process variable prediction unit 15C, a change speed calculation unit 15D, an abnormality sign detection unit 15E, and an alarm output unit 15F. be.

The target value trajectory calculation unit 15A approximates the transition of the target value of the target process variable included in the operation data 14A in the storage unit 14, calculates the target value at each monitoring timing, and stores the target value as target value trajectory data 14B in the storage unit. 14 has the function of saving. The target value of the target process amount contained in the operating data 14A indicates the transition from the start to the end of the batch process, but does not necessarily indicate a value synchronized with the monitoring timing. Therefore, the target value trajectory calculation unit 15A calculates the target value in advance for each monitoring timing by approximating with a known approximation method such as broken line approximation.

FIG. 2 is an example of generation of target value trajectory data. As shown in FIG. 2, when the target values at times t _i and t _i+1 are r _i and r _i+1 , the monitoring period τ included in the period from time t _i to time t _i+1 is The number j _max and the monitor timing t _b synchronized with the monitor period τ are obtained by the following equation (1).

Therefore, the target value r(t _b ) at the monitoring timing t _b , that is, the target value trajectory data 14B is obtained by the following equation (2) by linear approximation of the given target values r _i and r _i+1. be done. Note that when the target value r(t _b ) at the monitoring timing t _b is given as the target value r _i , the target value r _i may be used as the target value r(t _b ).

The data acquisition unit 15B sequentially acquires new target process quantities from the host device 20 and the controller 21 from the communication I/F unit 11 via the communication line L1, and stores them in the process quantity data 14C of the storage unit 14. Department.
When the new target process quantity is stored in the storage unit 14, the data acquisition unit 15B performs outlier correction processing and an exponential smoothing filter on the new target process quantity based on the target process quantities acquired so far. A smoothing process may be performed.

Based on the target value trajectory data 14B and the process amount data 14C, the process amount prediction unit 15C predicts the target process amount for each monitoring timing in the future interval from the time when the new target process amount is acquired to the end of the batch process. It is a processing unit that predicts a future prediction value and stores it in the prediction value data 14D of the storage unit 14. FIG.

Since the target value of the target process variable is known in the future interval, if the deviation prediction value, which is the deviation between the target value and the future prediction value in the future interval, is known, the future prediction value can be obtained by subtracting the deviation prediction value from the target value. be done.
The present invention first predicts deviation prediction values in the future interval from past deviations using a prediction algorithm, and calculates future prediction values by subtracting these deviation prediction values from target values. Moreover, the present invention uses an algorithm based on a known AR model (Auto-Regressive model) as a prediction algorithm.

An AR model can estimate the current value using only its past time series data. A value obtained by subtracting the target process variable y (t) from the target value r (t) at time t is defined as the deviation d (t), the order of the AR model is p (p is an integer of 2 or more), and the AR model When the ^model adjustment parameter is a=[a ₁ , _{a 2} , _.

When estimating the model adjustment parameter a, given the past time series {d(t), d(t-1), ..., d(t-2q)} (q>p) of the deviation including the current time, Using X representing the time series and the model adjustment parameter a, we obtain a simultaneous equation, d _pred =Xa, representing the estimated value time series by the AR model.
Next, solve the following optimization problem that minimizes the quadratic form of the estimate error to determine the model tuning parameters. At this time, in order to obtain a stable model adjustment parameter a even when the variation in the past time series is minute and contains a lot of noise, as shown in the following equation (4), the robustness adjustment parameter introduce λ.

As a result, the deviation prediction value d _pred (t+m) and the future prediction value y _pred (t+m), which are m (m is an integer equal to or greater than 1) in the future from the time t, are obtained by the following equation (5).

Here, the reason why the deviation prediction value is used when obtaining the future prediction value of the target process variable will be described. FIG. 3 shows an example of calculation of the future prediction value using the target process quantity, and FIG. 4 shows an example of monitor screen display of FIG. In FIGS. 3 and 4, the future prediction value was predicted with the monitoring cycle τ set to 2 minutes and the model adjustment parameter a in Equation (4) set to [p=40, q=120, λ=0.2].
In FIGS. 3 and 4, YH is historical value data indicating the transition of the past target process quantity y used to create the AR model, and YP is predicted value data indicating the transition of the future predicted value y _pred of the target process quantity. is. Further, YA is actual measurement value data indicating transition of the actual measurement value y of the target process variable, and SP is target value data indicating transition of the target value r of the target process variable.

As shown in FIGS. 3 and 4, when the future prediction value YP in the future interval F is predicted using the history value data YH of the target process variable, which is an absolute value, the target value data SP rises or falls in a ramp-like manner. The future prediction value YP deviates from the measured value data YA in the preceding and succeeding time zones, and a large error occurs in the future prediction value YP. This is because the target value data SP has changed, and the transition of the historical value data YH of the target process quantity used to create the AR model has largely deviated from the target value data SP. It has appeared.

On the other hand, FIG. 5 is a calculation example of the future prediction value using the deviation, and FIG. 6 is a monitoring screen display example of FIG. In FIGS. 5 and 6, the future prediction value was predicted with the monitoring cycle τ set to 2 minutes and the model adjustment parameter a in Equation (4) set to [p=40, q=120, λ=0.2].
5 and 6, DH is history value data indicating the transition of the past deviation d used to create the AR model, and DP is the future prediction value of the deviation, that is, the deviation prediction value indicating the transition of the deviation prediction value d _pred Data. Further, DA is the measured value data indicating the transition of the actual measured value d of the deviation, and SP is the target value data indicating the transition of the target value r of the target process variable. YP is predicted value data indicating the transition of the future predicted value y _pred of the target process variable, and YA is actual measurement data indicating the transition of the actual measurement value y of the target process variable.

As shown in FIG. 5, when the deviation prediction value DP is predicted using the deviation history value data DH, similar to FIG. DP diverges from the measured value data DA, and a large error occurs in the deviation predicted value DP. However, since the deviation, which is a relative value, always changes around zero, even if the target value data SP changes, the transition of the deviation history value data DH used to create the AR model does not change significantly. Therefore, the future prediction value YP of the target process quantity is not greatly affected, and a stable future prediction value YP can be obtained.

The change rate calculator 15D acquires the process variable data 14C and the predicted value data 14D from the storage unit 14, and calculates the change per unit time of the target process variable included in the process variable data 14C and the predicted value data 14D. It is a processing unit that calculates the change speed shown and stores it in the predicted value data 14D of the storage unit 14. FIG. Here, a moving average value of past N (N is an integer equal to or greater than 2) target process variables from time t is used to calculate the rate of change. Therefore, when the time t is a future time, the future prediction value is included in the target of the moving average process. In the following description of the calculation of the rate of change, the past value and the predicted future value of the target process variable may be referred to as the target process variable without distinction.

FIG. 7 is an explanatory diagram showing an example of conversion speed calculation. Assuming that the target process quantity at time t is y(t), the moving average y _mavg (t) of past N target process quantities from time t is obtained by the following equation (6).

Therefore, let τk be the sampling period of the target process variable, τ be the monitoring period for calculating the moving average y _mavg (t), Δt be the differential step time for calculating the rate of change, and Δt be the time difference corresponding to the differential step time Δt. Assuming n, the rate of change v(t) at time t is obtained by the following equation (7).

When calculating the rate of change, a weighted moving average (WMA) as shown in the following equation (8) may be used instead of the above simple moving average.

A smoothing process using a smoothing filter may be applied to the obtained rate of change to smooth the rate of change. At this time, by using a memory having a FILO (first in last out) structure as the stack memory, the smoothing process can be executed efficiently. If the monitoring cycle is τ, the differentiation step time for calculating the change speed is Δt, and the filter coefficient of the smoothing filter is α, the change speed smoothed value v _smooth is obtained by the following equation (9). Note that this is the filter coefficient α.

The abnormality sign detection unit 15E acquires the target value trajectory data 14B and the predicted value data 14D from the storage unit 14, and determines the preset allowable deviation for the target value r of the target value trajectory included in the target value trajectory data 14B. Calculate the allowable range rw by giving da, compare the future predicted value y _pred of the target process variable included in the predicted value data 14D with the allowable range rw, and generate in the future interval based on the obtained comparison result When a sign of possible abnormality is detected, the deviation between the future prediction value y _pred and the target value r is greater than a predetermined allowable deviation, and the future prediction value y _pred deviates from the allowable range rw, an alarm is issued to the alarm output unit 15F. This is the processing unit for output.

The allowable range rw(t+m) is defined by the allowable range upper limit rwh(t+m)=r(t+m)+da obtained by adding the allowable deviation da to the target value r(t+m), and the allowable range rwh(t+m)=r(t+m)+da obtained by subtracting the allowable deviation da from the target value r(t+m). The lower range limit is defined as rwl(t+m)=r(t+m)-da.
When the future predicted value y _pred (t+m) exceeds the allowable upper limit rwh (t+m), or when the future predicted value y _pred (t+m) falls below the allowable lower limit rwl (t+m), an alarm related to the process variable is issued. output.

Further, the abnormality portent detection unit 15E compares the change speed prediction value v _pred included in the prediction value data 14D with a preset threshold value vth, and based on the obtained comparison result, a It is a processing unit that detects a sign of possible abnormality and causes the alarm output unit 15F to output an alarm when the change speed prediction value _vpred exceeds the threshold value _vth .

The warning output unit 15F has a function of outputting an alarm indicating that a sign of an abnormality that can occur in the future section has been detected in response to an instruction from the abnormality sign detection unit 15E. The alarm output from the alarm output unit 15F may be displayed on the screen by the screen display unit 13, or may be notified from the communication I/F unit 11 to the host device 20 via the communication line L1.

[Operation of this embodiment]
Next, operation of the operation monitoring device 10 according to the present embodiment will be described with reference to FIG. FIG. 8 is a flowchart showing operation monitoring processing.
The operation monitoring device 10 executes the operation monitoring process of FIG. 8 in response to the start of the batch process. It is assumed that the operation data 14A is stored in the storage unit 14 when the operation start process is executed. In addition, it is assumed that the monitoring timing and the amount of processes to be monitored are set in advance.

First, the target value trajectory calculation unit 15A acquires the operation data 14A from the storage unit 14, and approximates the transition of the target value of the target process variable included in the operation data 14A by a known approximation method such as polygonal line approximation. , the target value r is calculated at each monitoring timing and stored in the storage unit 14 as target value trajectory data 14B (step S100).

Next, the data acquisition unit 15B confirms the arrival of the monitoring timing (step S101), and waits until the arrival of the monitoring timing (step S101: NO).
When the arrival of the monitoring timing is confirmed (step S101: YES), the data acquisition unit 15B receives a new target process quantity y from the host device 20 or the controller 21 from the communication I/F unit 11 via the communication line L1. It is acquired and stored in the process variable data 14C of the storage unit 14 (step S110).

Next, the process variable prediction unit 15C generates an AR model based on the target value trajectory data 14B and the process variable data 14C acquired from the storage unit 14, and terminates the batch process when a new target process variable y is acquired. A deviation prediction value d _pred indicating a deviation between a new target process variable y and a target value r is predicted (step S111), and the obtained deviation prediction value d _pred and the target value r, the future prediction value _ypred , which is the prediction value of the target process quantity, is calculated for each monitoring timing in the future section and stored in the prediction value data 14D of the storage unit 14 (step S112).

Next, based on the process variable data 14C and the predicted value data 14D acquired from the storage unit 14, the change rate calculator 15D calculates the change in the future predicted value y _pred included in the process variable data 14C and the predicted value data 14D. A speed prediction value v _pred is calculated and stored in the prediction value data 14D of the storage unit 14 (step S113).

After that, the abnormality sign detection unit 15E acquires the process variable data 14C, the target value trajectory data 14B, and the predicted value data 14D from the storage unit 14, and for each monitoring timing, the process variable y, A monitoring situation screen is generated by plotting the target value r, the allowable range rw, the future prediction value y _pred , and the change speed prediction value v _pred in chronological order, and is displayed on the screen display unit 13 (step S114). At this time, the allowable range rw is calculated by giving a preset allowable deviation da to the target value r.

Subsequently, the abnormality sign detection unit 15E checks whether the future prediction value y _pred of the target process quantity deviates from the allowable range rw (step S115), and if it deviates (step S115: YES ), the alarm output unit 15F outputs an alarm indicating detection of an anomaly sign of the target process variable (step S116), and if there is no deviation (step S115: NO), the process proceeds to step S117.

Further, the abnormality portent detection unit 15E checks whether or not the change speed prediction value v _pred of the target process quantity exceeds a preset threshold value v _th (step S117), and if it exceeds (step S117 : YES), the alarm output unit 15F outputs an alarm regarding the change rate prediction value v _pred of the target process variable (step S118), and if there is no deviation (step S117: NO), the process proceeds to step S102.

After that, the data acquisition unit 15B confirms whether or not the operation of the batch process has ended (step S102), and if it is in operation (step S102: NO), the process returns to step S101.
On the other hand, if the operation is finished (step S102: YES), the series of operation monitoring processes is finished.

[Example of operation]
Next, an operation example of the operation monitoring device 10 according to the present embodiment will be described with reference to FIGS. 9 and 10. FIG. FIG. 9 is an example of a monitor screen display showing changes in deviation. FIG. 10 is an example of a monitor screen display showing changes in future prediction values.
9 and 10, the batch process and the operation monitoring operation were simulated with a certain reaction temperature in the batch process as the target process amount, and when predictive monitoring was performed at monitoring time t 150 minutes after the start of the batch process, An example of the monitor screen display obtained is shown.

In FIG. 9, DH is data indicating the transition of the past deviation d used for the deviation prediction at the monitoring time t, DP is data indicating the transition of the deviation prediction value d _pred predicted at the monitoring time t, and DA is data indicating the transition of the deviation d between the target process variable y and its target value r.
In FIG. 10, YA is data indicating the transition of the target process quantity y, and YP is data indicating the transition of the future predicted value y _pred of the target process quantity. Further, SP is target value data indicating the transition of the target value r, and RW is data indicating the transition of the allowable range rw.

In this simulation, the monitoring cycle τ was set to 2 minutes. Further, the time length of the future interval is set to 60 minutes, and the future prediction value y _pred is predicted and monitored for the future interval F from the monitoring time t to the monitoring end time _te of the 210th minute from the start of the batch process. . Note that the aforementioned p, q, and λ, which are the model adjustment parameters of the AR model, were set to 10, 20, and 0.2, respectively.

According to the examples of FIGS. 9 and 10, the future prediction value y _pred gradually deviates from the target value r after the prediction time t at 150 minutes from the start of the batch process, and reaches the allowable future time ta at 180 minutes. It can be seen that the upper limit of the range rw is exceeded. This means that if the target process variable y changes as it is, it will be detected at the predicted time t, 30 minutes earlier, that the upper limit of the allowable range rw will be exceeded at the 180th future time ta. there is

[Effects of this embodiment]
As described above, in the present embodiment, the process amount prediction unit 15C predicts the future predicted value of the process amount at the future monitoring timing included in the future interval based on the time-series data of the process amount obtained in the past. Then, the anomaly sign detection unit 15E detects an anomaly sign that can occur in the future interval based on the future prediction value and the target value at the future monitoring timing. Further, when the deviation between the future predicted value and the target value is larger than a predetermined allowable deviation, an alarm is output indicating that a sign of abnormality that may occur in the future interval has been detected.

As a result, when the future predicted value of the process variable deviates from the target value of the target value trajectory by more than the allowable deviation, or when the rate of change of the future predicted value deviates from the allowable range, it is immediately detected as an anomaly sign. Furthermore, an alarm will be output. Therefore, for each monitoring timing, it is possible to confirm the presence or absence of an abnormality that may occur in the future interval from the monitoring timing, and it is possible to accurately and early detect a sign of abnormality in the batch process.

Further, in the present embodiment, the change speed calculation unit 15D calculates the change speed of the future predicted value of the process quantity at the future monitoring timing, and the abnormality sign detection unit 15E is preset with the obtained change speed. It is designed to detect a sign of abnormality that may occur in the future interval based on the obtained comparison result. In addition, when the rate of change exceeds the threshold value, an alarm is output indicating that a sign of abnormality that may occur in the future section has been detected.

This makes it possible to accurately and early detect signs of abnormalities in batch processes that are likely to cause serious accidents, such as sudden rises in reaction temperature. It should be noted that the abnormality sign detection unit 15E may not detect an abnormality sign based on the comparison between the future prediction value and the target value of the target value trajectory, but may detect only the abnormality sign based on the rate of change.

Further, in the present embodiment, the process variable prediction unit 15C predicts the deviation prediction value of the deviation at the future monitoring timing based on the deviation between the process variable obtained in the past and the target value for the process variable, By subtracting the predicted deviation value from the target value of the target value trajectory at each future monitoring timing, the future predicted value of the process variable at that future monitoring timing may be calculated.

As a result, since the deviation is data that fluctuates up and down around a constant value, it is possible to predict with high accuracy using an autoregressive model such as an AR model. As a result, it is possible to obtain more stable future prediction values than when predicting future prediction values based on process quantities obtained in the past. can be detected.

[Expansion of Embodiment]
Although the present invention has been described with reference to the embodiments, the present invention is not limited to the above embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention. In addition, each embodiment can be implemented in any combination within a non-contradictory range.

DESCRIPTION OF SYMBOLS 10... Operation monitoring apparatus 11... Communication I/F part 12... Operation input part 13... Screen display part 14... Storage part 14A... Operation data 14B... Target value trajectory data 14C... Process quantity data 14D ... Predicted value data 15 ... Arithmetic processing unit 15A ... Target value trajectory calculator 15B ... Data acquisition unit 15C ... Process variable prediction unit 15D ... Change speed calculation unit 15E ... Abnormal sign detection unit 15F ... Alarm output Part 20... Host device 21... Controller 22... Sensor 23... Actuator L1, L2... Communication line.

Claims (7)

An operation monitoring device that monitors the operation status of the batch process at each monitoring timing based on the process amount to be monitored obtained from the batch process,
a target value trajectory calculation unit that calculates a target value trajectory indicating the target value at the monitoring timing based on the preset transition of the target value related to the process variable;
a process variable prediction unit that predicts a future predicted value of the process variable at a future monitoring timing included in a future interval based on time-series data of the process variable obtained in the past;
an anomaly sign detection unit that detects a sign of an anomaly that may occur in the future section based on the future predicted value at the future monitoring timing and the target value indicated by the target value trajectory ,
The process variable prediction unit calculates a deviation between the process variable and the target value obtained at each past monitoring timing based on the time-series data of the process variable obtained in the past and the transition of the target value indicated by the target value trajectory. for each future monitoring timing, predicting the deviation at the future monitoring timing, and subtracting the deviation from the target value at the future monitoring timing among the target values indicated by the target value trajectory By calculating the future prediction value at the future monitoring timing,
A driving monitoring device characterized by: In the operation monitoring device according to claim 1,
further comprising a change speed calculation unit that calculates a change speed indicating a change amount per unit time of the future prediction value at the future monitoring timing,
The anomaly sign detection unit compares the rate of change with a preset threshold value at each of the future monitoring timings, and detects an anomaly sign that may occur in the future interval based on the obtained comparison result. A driving monitoring device characterized by detecting An operation monitoring device that monitors the operation status of the batch process at each monitoring timing based on the process amount to be monitored obtained from the batch process,
a target value trajectory calculation unit that calculates a target value trajectory indicating the target value at the monitoring timing based on the preset transition of the target value related to the process variable;
a process variable prediction unit that predicts a future predicted value of the process variable at a future monitoring timing included in a future interval based on time-series data of the process variable obtained in the past;
a change speed calculation unit that calculates a change speed indicating a change amount per unit time of the future prediction value at the future monitoring timing;
An anomaly sign detection unit that compares the change rate with a preset threshold value at each of the future monitoring timings, and detects an anomaly sign that may occur in the future interval based on the obtained comparison result. and _
The process variable prediction unit calculates a deviation between the process variable and the target value obtained at each past monitoring timing based on the time-series data of the process variable obtained in the past and the transition of the target value indicated by the target value trajectory. for each future monitoring timing, predicting the deviation at the future monitoring timing, and subtracting the deviation from the target value at the future monitoring timing among the target values indicated by the target value trajectory By calculating the future prediction value at the future monitoring timing,
A driving monitoring device characterized by: In the operation monitoring device according to claim 1 or claim 2,
Further comprising an alarm output unit that outputs an alarm indicating that a sign of abnormality that may occur in the future interval has been detected,
The abnormality sign detection unit causes the alarm output unit to output an alarm when a deviation between the future predicted value and the target value at the future monitoring timing is greater than a predetermined allowable deviation. . In the operation monitoring device according to claim 2 or 3,
Further comprising an alarm output unit that outputs an alarm indicating that a sign of abnormality that may occur in the future interval has been detected,
The operation monitoring device, wherein the abnormality sign detection unit causes the alarm output unit to output an alarm when the rate of change exceeds the threshold value. An operation monitoring device comprising a target value trajectory calculation unit, a process amount prediction unit, and an abnormality sign detection unit, and monitoring the operation status of the batch process at each monitoring timing based on the process amount to be monitored obtained from the batch process. An operation monitoring method used in
a target value trajectory calculation step in which the target value trajectory calculation unit calculates a target value trajectory indicating the target value at the monitoring timing based on the transition of the target value related to the process variable set in advance;
The process variable prediction unit predicts a future predicted value of the process variable at each future monitoring timing included in the future interval based on the time-series data of the process variable obtained in the past. a process quantity prediction step for
The anomaly sign detection unit detects, for each of the future monitoring timings, a sign of an anomaly that may occur in the future interval based on the future predicted value at the future monitoring timing and the target value indicated by the target value trajectory. and an anomaly sign detection step for detecting
In the process variable prediction step, the process variable prediction unit uses the time-series data of the process variable obtained in the past and the transition of the target value indicated by the target value trajectory to determine the process value obtained at each past monitoring timing. Based on the deviation between the amount and the target value, the deviation at the future monitoring timing is predicted for each of the future monitoring timings, and the target value indicated by the target value trajectory at the future monitoring timing is predicted. calculating the future prediction value at the future monitoring timing by subtracting the deviation from the target value;
A driving monitoring method characterized by: Equipped with a target value trajectory calculation unit, a process amount prediction unit, a rate of change calculation unit, and an anomaly sign detection unit, the operation status of the batch process is monitored at each monitoring timing based on the process amount to be monitored obtained from the batch process. An operation monitoring method used in an operation monitoring device for monitoring,
a target value trajectory calculation step in which the target value trajectory calculation unit calculates a target value trajectory indicating the target value at the monitoring timing based on the transition of the target value related to the process variable set in advance;
The process variable prediction unit predicts a future predicted value of the process variable at each future monitoring timing included in the future interval based on the time-series data of the process variable obtained in the past. a process quantity prediction step for
a change speed calculation step in which the change speed calculation unit calculates, for each of the future monitoring timings, a change speed indicating an amount of change per unit time of the future prediction value at the future monitoring timing;
The anomaly sign detection unit compares the rate of change with a preset threshold value at each of the future monitoring timings, and predicts an anomaly that may occur in the future interval based on the obtained comparison result. and an anomaly sign detection step for detecting
In the process variable prediction step, the process variable prediction unit uses the time-series data of the process variable obtained in the past and the transition of the target value indicated by the target value trajectory to determine the process value obtained at each past monitoring timing. Based on the deviation between the amount and the target value, the deviation at the future monitoring timing is predicted for each of the future monitoring timings, and the target value indicated by the target value trajectory at the future monitoring timing is predicted. calculating the future prediction value at the future monitoring timing by subtracting the deviation from the target value;
A driving monitoring method characterized by:

Images