JPH04366742A - Plant monitoring device and its method - Google Patents

Abstract

PURPOSE:To surely find out the abnormalities of a plant in an early stage without depending on judgement made by an operator. CONSTITUTION:The historical data base 6 is provided, which stores both the time series data of set values for a plant from the past to the present and the time series data of the operation quantities of the plant from the past to the present. The time series data within the historical data base 6 is taken out by a process analyzing mechanism 7, and the coefficient of correlation between each set value and each operation quantity is computed, and it is judged based on the coefficient of correlation whether or not the plant is in operation normally, so that the result is thereby indicated on a CRT 9.

Description

[Detailed description of the invention]

0001

[Industrial Application Field] The present invention compares measured values from various sensors installed in a plant with preset set values, and operates the plant so as to reduce the deviation between the measured values and the set values. The present invention relates to a plant monitoring device and method for monitoring a plant by controlling.

0002

[Prior Art] When controlling a plant in a continuous process, a controller is used to control an object to be controlled, as described in Japanese Patent Application Laid-Open No. 1-25600. The measured value (current value of the controlled object) PV, set value (target value of the controlled object) SV, and manipulated variable (amount of operation instruction to the controlled object) MV are handled as data. While monitoring these data, the operator visually checks whether or not an abnormality has occurred in the plant, the controlled object, etc. using the sensor data, trend graph, and historical data trend graph displayed by the control system. When obtaining a trend graph or a historical data trend graph, mathematical analysis methods such as correlation coefficients and linear regression are generally used.

0003

[Problem to be solved by the invention] Looking at recent trends in the personnel allocation of operators involved in plant operation, the number of new operators assigned is decreasing year by year. Many companies are cutting back on staff. In addition, due to the aging of operators, the trend of decreasing the number of operators is becoming even more pronounced.

[0004] Under these circumstances, the amount of plant monitoring data that one plant operator is responsible for has become enormous. Specifically, in a plant with about 3,000 sensor data points, if there are 3 to 4 operators, each worker will receive 80
You have to monitor about 0 to 1000 points.

[0005] Furthermore, if the data to be monitored changes extremely slowly, unless the operator constantly monitors the data for a long period of time, it will be difficult to detect the characteristics (the gradual progress of the abnormality). However, with the notification function that outputs an alarm when the alarm setting value is exceeded, the degree of abnormality varies greatly.

[0006] For this reason, the speed at which abnormal signs can be detected is
It can be said that it influences the response method, response time, and difficulty of response for abnormality avoidance, but conventional systems do not have such a monitoring function, so everything must depend on the operator's monitoring ability based on experience.

[0007] It has been found that when past trend data at the time of an abnormality occurrence is investigated, it is often possible to read the state of transition to the abnormal state from the data. However, in the above-mentioned conventional technology, the safe operation area, alarm area, and out-of-tolerance area determined by the screen display of a plurality of correlated process quantities and the correlation are displayed. The method of notifying operators of alarms, which is decreasing in number, has the disadvantage that the burden on the operators increases significantly because the operators have to actively refer to the screen.

[0008] Furthermore, the correlation formula in the prior art cannot be applied to all control loops, but only to limited process data; therefore, the correlation formula can only be applied to process data for which the correlation formula is known. It also has the disadvantage that the scope of application is very narrow.

The purpose of the present invention is to analyze historical data stored in a computer so that signs of plant abnormality can be detected at an early stage without placing a burden on operators, and to have a wide range of applicability. An object of the present invention is to provide a plant monitoring device and method.

0010

[Means for Solving the Problems] In order to achieve the above object, the present invention captures a measured value from a sensor installed in a plant, compares the measured value with a preset setting value, and measures the measured value. In a plant monitoring device that monitors a plant by controlling a manipulated variable of the plant so that a deviation between a value and a set value is small, a set value storage means stores time-series data of the set value from the past to the present. and a manipulated variable storage means for storing time-series data of the plant manipulated variables from the past to the present, and time-series data of the set values stored in the set value storage means and manipulated variables stored in the manipulated variable storage means. Calculate the correlation with the time series data of
The apparatus is equipped with a determination means for determining whether the plant is normal or abnormal based on the calculation result, and an output means for outputting the determination result.

[0011] Furthermore, the present invention takes in measured values from sensors installed in a plant, compares the measured values with preset set values, and adjusts the plant so that the deviation between the measured values and the set values is small. By controlling the amount of operation of
In a plant monitoring device that monitors a plant, a set value storage means stores time-series data of the set values from the past to the present, and a manipulated variable memory stores time-series data of the plant manipulated variables from the past to the present. and calculates the correlation between the time-series data of the set values stored in the set value storage means and the time-series data of the manipulated variables stored in the manipulated variable storage means, and from the calculation results, it is possible to detect abnormalities in the plant in the future. It is equipped with a prediction means for predicting what will happen, and an output means for outputting the prediction result.

[0012] Furthermore, the present invention is that one of the above-mentioned plant monitoring devices is installed in a flow rate control device, a gas process, or a calorific process.

Furthermore, the present invention takes in measured values from sensors installed in the plant, compares the measured values with preset set values, and adjusts the plant so that the deviation between the measured values and the set values is small. In a plant monitoring method for monitoring a plant by controlling the manipulated variable of The correlation between the stored set value time series data and the manipulated variable time series data is calculated, and the normality or abnormality of the plant is determined based on the calculation result.

Furthermore, the present invention captures a measured value from a sensor installed in a plant, compares the measured value with a preset setting value, and makes the deviation between the measured value and the set value small. In a plant monitoring method for monitoring a plant by controlling the manipulated variable of the plant, time-series data of the set value from the past to the present is stored, and time-series data of the plant manipulated variable from the past to the present is stored. is stored, the correlation between the stored time-series data of set values and the time-series data of manipulated variables is calculated, and future abnormalities in the plant are predicted from the calculation results.

[0015]

[Operation] Signs of abnormality in the control loop can be detected as follows, especially in the steady state of a continuous process. That is, in a steady state, the relationship between the set value SV of the control loop and the manipulated variable MV is that unless the set value SV is changed, the manipulated variable MV is
remains constant. In other words, by referring to the time-series changes in the set value SV, we monitor whether it is in a steady state or not, and if it is determined that it is constant (no change), the manipulated variable M at this time
Refer to the time-series changes in V and see the change trends.

[0016] Furthermore, when referring to this time-series data, index data for determining the presence or absence of a correlation is obtained by determining the correlation coefficient between the set value SV and the manipulated variable MV using the time-series data. By storing this data as historical data, it becomes possible to refer to changes in the correlation coefficient over time.

[0017] Furthermore, by periodically executing the monitoring function and analyzing past time-series data of previously registered process data, a computer can perform the same function as a human looking at a display screen and checking data changes. If an abnormality occurs in the plant, it is possible to automatically and reliably notify operators of the abnormality.

Next, details of the present invention will be explained using FIG. 5. The present invention provides process variables (
It is applied to control loops having information on set values SV, manipulated variables MV, and measured values PV, including temperature control, flow rate control, position control, rotation speed control, etc. In particular, among the above information, the set value SV and the manipulated variable MV are used as past time series data.

The manipulated variable MV varies with respect to the controlled object in order to satisfy the set value SV (this variation refers to a command to open the valve to what extent in the case of flow rate control based on the valve opening degree, or a command to open the valve to a certain degree in temperature control). However, in a steady state, the manipulated variable MV remains constant unless the set value SV is changed.

Here, a state in which the set value SV does not fluctuate is called a steady state, and a state in which the set value SV fluctuates is called a transient state. Furthermore, in plant control, since SV=PV, the controller controls the manipulated variable MV, but it is possible to differentiate between when the controller is working and when it is not working.

A feature of the present invention is that it is checked whether the controller is working more than necessary in a steady state. a. When the manipulated variable MV does not fluctuate in a steady state (however, here it is limited to not fluctuating above a certain specified value), the relationship of SV = PV is maintained, so the plant control is performed normally. It is thought that there are. b. When the manipulated variable MV fluctuates in steady state, SV=PV
Since the relationship is not maintained, the controller
If the manipulated variable MV is manipulated to return to the relationship of V, and it is considered that a disturbance has occurred, and if the manipulated variable MV continues to fluctuate abnormally, it is considered that there is an abnormal cause such as equipment failure.

Next, the process of calculating the correlation coefficient will be described. The process of calculating the correlation coefficient can be calculated using the following formula.

[0023]

[Math 1]

[0024] The correlation coefficient r is an index showing the degree of interrelationship between two variables (set value SV, manipulated variable MV), and when the two variables are independent, r = 0, and the two variables are a perfect linear function. If there is a relationship, r=±1. Then, as the value approaches from around 0 to around ±1, it is possible to detect whether the causal relationship between the two variables is stronger.

Therefore, by monitoring whether r<the specified value, it is possible to monitor whether the causal relationship between the set value SV and the manipulated variable MV is maintained. Note that the value of r≧specified value differs depending on the process and the function of the control loop, so it will not be discussed here.

[0026]

EXAMPLES The present invention will be explained in detail below according to examples. When considering an automatic control system model for the basic control loop configuration in a plant control system, during steady operation, excluding startup and shutdown, chemical plant control is overwhelmingly dominated by setpoint control. There are feedback loop and feed-forward loop control systems as forms for performing set value control. Furthermore, PID controllers are widely used as control systems for these feedback loops and feed-forward loops.

For example, in feedback control, the set value SV is given as a reference input, and the deviation between the measured value PV, which is the feedback amount of the control loop, and the set value SV is given to the adjustment section as an operating signal, and when this deviation becomes 0, The manipulated variable MV is determined through the controller (operation unit) so that the controlled object is controlled. At this time, in the controlled object, the detection unit collects the measured value PV and provides it to the controller. Here, the controller is a computer control or an analog controller, the operation part is an on-off valve, etc., and the detection part is a variety of sensors.

Next, the application of the plant monitoring device of the present invention will be described by citing specific processes.

a. Flow Control in Capacitive Process The capacitive process described here applies to cases such as controlling the water level in a tank. Specifically, consider the following model. In a tank with outflow force, if the set value SV is the control target water level in the tank, the measured value PV is the current water level in the tank, and the manipulated variable MV is the opening/closing degree of the inflow control valve that controls the flow rate to the tank, then In a steady state (when the plant is normal), if the outflow flow from the tank is constant, the deviation between the set value SV and the measured value PV is constant, so the inflow flow per unit time is constant, and the inflow The degree of opening and closing of the control valve, that is, the manipulated variable MV remains constant. In other words, it can be said that the manipulated variable MV is constant unless the set value SV, which is the control target, is changed. However, it should be noted here that there are steady states and transient states in process control, so the above applies only to steady states (here, steady state means that the process does not fluctuate). A transient state refers to a state in which the process is stable without any changes, and a transient state refers to a state that exhibits transient process behavior due to process fluctuations, and has a different meaning from the steady state described above).

Next, consider the case where an abnormal condition such as a liquid leak occurs inside the tank or in the flow path. Before the abnormality occurs, the set value SV and the manipulated variable M are the same as in the steady state described above.
Both V remain constant, but when a disturbance such as a liquid leak occurs, the outflow flow rate starts to gradually increase compared to before the abnormality occurs. In other words, the measured value PV of the current water level gradually decreases even though the set value SV, which is the control target, is not changed. In this case, the controller increases the opening degree of the inflow control valve, that is, the control amount PV. Considering the passage of time, liquid leakage progresses gradually over a long period of time, so deviations from the SV-PV control target value are not recognized as errors when visually judged by humans. There are many. However, this deviation can be fully confirmed after a sufficient period of time has passed, but even if it is confirmed at this time, it is often considered a plant abnormality and is in a dangerous state. . Therefore,
If the plant monitoring device of the present invention is applied to such a capacity process, problems in the process can be quickly detected and plant abnormalities can be avoided in advance.

b. Gas Process In the gas process as well, as in the case of the above-mentioned volumetric process, pressure drops such as gas leaks can be detected early in the pressure control loop. Furthermore, in the case of a complex control loop, it is possible to monitor a wider range of process data by monitoring using calculation results such as material balance rather than by monitoring a single pressure value.

c. Calorie process In the case of a calorific process, by combining it with calculation results such as heat balance, it becomes possible to monitor a wider range of processes than capacity processes or gas processes. In this case, data necessary for heat balance calculation (heat amount, flow rate, etc.) is subject to abnormality monitoring.

In order to control each of the above-mentioned processes, the present invention deals with this by applying correlation processing of process analysis processing. The correlation coefficient is expressed in the range -1≦C≦+1, and has meaning depending on the sign and the size of the numerical value. The sign represents positive correlation and negative correlation, and the magnitude of the numerical value represents the strength of the degree of connection between the two correlated data. Further, when C=0, it indicates that there is no correlation between these two data.

FIG. 1 is a block diagram showing the operating environment of the plant control system of the present invention. The plant control device of the present invention includes a process input/output device 1, a data collection mechanism 2, a man-machine mechanism 3, a process control processing mechanism 4, a plant database 5, a historical database 6, and a process analysis mechanism 7. A target plant 8 is connected to the process input/output device 1, and an input/output device 9 is connected to the man-machine mechanism 3.

The historical database 6 collects and stores various data including process data in chronological order from time to time. And man-machine mechanism 3
When information for activating the process analysis mechanism 7 (data name, analysis data collection start time, number of pieces of analysis data, etc.) is provided from the program or another program, analysis processing is executed under the specified conditions. The analysis results are correlated as state variables.
It can be displayed in Japanese using the words "Yes/No". Furthermore, in the analysis results, it is also possible to output and display the above-mentioned state variables as management codes.

Next, the processing in the process analysis mechanism 7 will be described. The processing in the process analysis mechanism 7 is (1
) Historical data collection processing 1, (2) Correlation processing, (
It consists of 3) historical data registration processing, (4) historical data collection processing 2, (5) linear regression processing, and (6) state variable conversion processing. Each of the above processes will be explained in detail below.

(1) Historical data collection processing 1 Collect target historical data based on the analytical data collection start time and data number given as startup information. The analysis data collection start time can be specified as either an absolute time or a relative time.

(2) Correlation Process Correlation coefficient (correlation) is calculated for the time series data collected by the above historical data collection process 1.
n coefficient).

(3) Historical data registration process The correlation coefficient obtained by the above correlation process is registered in historical data.

(4) Historical data collection process 2 In order to examine the time-series transition of the correlation coefficients registered by the above historical data registration process, related time-series correlation coefficient data is extracted from the historical database 6.

(5) Linear regression processing For the time-series data of the correlation coefficients collected in the above historical data collection process 2, linear regression processing is executed to obtain an approximate straight line, and the slope thereof is further calculated. Analyze the transition (fluctuation state) of the correlation coefficient.

(6) State variable conversion processing Based on the slope information of the approximate straight line obtained in the above linear regression processing, the state variables are replaced with state variables that are understandable to humans or normally used by operators.

Next, a specific example of correlation processing will be explained. FIG. 2 is a flowchart of correlation processing. First, in step 21, the set value S is determined from the historical database.
The time series data of V is collected, and in step 22 it is determined whether there is a step change in the time series data of the set value SV, that is, there is a change in the set value SV. If there is no change, the process ends, and if there is a change, time-series data of the manipulated variable MV is collected from the historical database in step 23. Next, in step 24, the set value S
Calculate the standard deviation of the set value SV from the time series data of V,
In step 25, the standard deviation of the manipulated variable MV is calculated from the time series data of the manipulated variable MV. Furthermore, step 24
Using the standard deviation of the set value SV calculated in step 25 and the standard deviation of the manipulated variable MV calculated in step 25, step 26
Calculate the correlation coefficient r. And step 27
It is determined whether the correlation coefficient r calculated in step 2 is less than a specified value, and if it is less than the specified value, a message indicating an abnormal correlation is output in step 28, and if it is greater than or equal to the specified value, the correlation state is determined to be normal in step 29. It is processed.

(First Embodiment) This embodiment is a process alarm output process in a process control system.
Conventional instrument alarm high-error, low-
Error, deviation error, rate of change error, control loop set value/manipulated variable (SV/MV) abnormality monitoring alarm is added, and is called an intelligent alarm.

The SV/MV abnormality monitoring alarm patrols the process monitoring process using the above-mentioned correlation analysis process.
For the set value SV and manipulated variable MV of Tag), instantaneous value alarms such as "Manipulated variable MV is increasing even though set value SV remains unchanged" and "Managed variable MV is decreasing even though set value SV remains unchanged" It outputs an alarm that takes past data into account.

In this embodiment, correlation processing is performed between the set value SV and the manipulated variable MV. However, when SV is constant in a steady state and the controller is working normally, the set value SV and the manipulated variable MV are Since the quantities MV are constant values,
There is no correlation (strictly speaking, this means that since there is no change in either data, it is not possible to determine whether there is a correlation during processing).

[0047] Next, when SV = constant, when the controller operates incorrectly due to process abnormality or due to disturbance, SV = constant and MV = fluctuating, so correlation processing is possible unlike the above. becomes. However, here as well, there may be a positive correlation value depending on the characteristics of the target process, so it is not possible to make a definitive statement about the sign. Furthermore, as the degree of abnormality increases, the fluctuation range of the manipulated variable MV increases, so this can also be captured by the correlation coefficient.

As shown in FIG. 3, the plant monitoring device of this embodiment includes a process input/output device 1, a data collection mechanism 2,
It is composed of a man-machine mechanism 3, a process control processing mechanism 4, a plant database 5, a historical database 6, a process analysis mechanism 7, an input/output device 9, a data access management mechanism 10, and a patrol monitoring mechanism 11. The process input/output device 1 is connected to the target plant 8.

Data from various sensors installed in the target plant 8 is collected by the data collection mechanism 2 via the process input/output device 1, and after being subjected to arithmetic processing by the process control processing mechanism 4, the data is stored in the plant database 5. will be written to. The data written to the plant database 5 is periodically transferred to the historical database 6.
The data is transferred to the historical database 6 and stored as historical data for a long period of time. The contents of the plant database 5 and the contents of the historical database 6 are as follows.
Through the man-machine mechanism 3, it can be easily referenced by the input/output device 9. Further, some data may be output to the target plant 81 via the process control processing mechanism 4. The process control processing mechanism 4 also checks the rationality of the collected process data, and at this time checks whether the process data is within the measurement range or exceeds the alarm output setting value (specified value). In addition to checking, the deviation and rate of change between the process data and the previously collected process data are also calculated, and checks are performed to see if each exceeds the alarm alarm output value. Then, an alarm warning is output to the man-machine mechanism 3 according to the result.

The above-mentioned correlation processing is stored in the process analysis mechanism 7, and the processing is executed according to analysis parameters specified in advance. At this time, in order to perform SV/MV monitoring, the set value SV and operation amount MV of the process data, analysis time (how many minutes ago historical data will be used as analysis data), and whether or not to display man-machine display are specified. be done.

The data access management mechanism 10 manages access to the plant database 5 and historical database 6.

The patrol monitoring mechanism 11 manages the above-mentioned process analysis processing contents that have been registered in advance, periodically issues execution requests to the process analysis mechanism 7, executes the analysis processing, and obtains the analysis results. Then, the result is displayed as an alarm on a graphic CRT, which is one of the input/output devices 9, via the man-machine mechanism 3.

(Second Embodiment) This embodiment is a system in which a knowledge processing mechanism is added to a process control system.
This is a knowledge processing system that allows the results obtained by the process analysis mechanism to be used as inference data in the inference executed within the knowledge processing mechanism, and the words (state variables) used by the operator in inference can be used as they are. It is possible to use it.

In the actual rule description,
IF (FIC1000 @loop state is pipe occlusion state) THEN
:
: I
F(FIC1000 @loop state is
There is fluid leakage inside the pipe) THEN
:
: Since it becomes possible to use the expressions that operators normally think of, knowledge processing, rule input, and maintenance are simplified.
It can be executed more efficiently than before.

As shown in FIG. 4, the plant monitoring device of this embodiment includes a process data input/output device 1, a data collection mechanism 2, a man-machine mechanism 3, a process control processing device 4, a plant database 5, a historical database 6,
Process analysis mechanism 7, input/output device 9, database access mechanism 10, inference mechanism 12, and knowledge database 1
It consists of 3. And process input/output device 1
is connected to the target plant 8.

Data from various sensors installed in the target plant 8 is collected by the data collection mechanism 2 via the process input/output device 1 and written to the plant database 5, as in the first embodiment. . In addition, the same applies to the case where data is collected from another computer system via a data communication device and written to the plant database 5. Data in the plant database 5 is periodically transferred to a historical database 6 and stored in this historical database 6 as historical data for a long period of time. The contents of the plant database 5 and the contents of the historical database 6 can be easily referenced through the man-machine mechanism 3. The process control processing mechanism 4 performs arithmetic processing on the collected process data and processes the data. Furthermore, some data may be output to the target plant 8 via the process input/output device.

[0057] In the knowledge database 13, IF...TH
Production rules written in the EN... format are registered, and the production rules registered in the knowledge database 13 are executed by the inference mechanism 12. When a production rule is executed, if a macro name that requests execution to the process analysis mechanism is written in the THEN portion of the production rule, execution of the production rule causes an execution request to be made to the process analysis mechanism. Based on this request, the process analysis mechanism performs correlation processing.

IF
: THEN(send apos.
The analysis processing result of apos_correl(...)) is
This is reflected in the frame that handles fact-based knowledge in the knowledge database. In the reflected state, the production rule expressed by the above-mentioned state variables will be executed.

[0059] Furthermore, the analysis processing results executed by inference are as follows:
It can be referenced by the input/output device 9 via the man-machine mechanism 3.

[0060]

[Effects of the Invention] As explained above, according to the present invention,
It has the following effects.

a. Since computers can be used to make judgments about process status that used to be made by humans, it becomes possible to automate plant monitoring.

b. Since it is possible to express knowledge in the same way as humans memorize it, it is possible to construct a plant monitoring system that is easy for humans to understand.

c. It is now possible to reliably and accurately report process alarms that were easy to overlook in the past, and it is now possible to output process alarms that were not possible in the past.

d. Since it is possible to grasp the long-term status of a plant, it is also possible to reliably capture slow movements that humans would not notice.

e. Knowledge processing can be equipped with a status understanding function, allowing optimization of plant monitoring.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the operating environment of a plant monitoring device of the present invention.

FIG. 2 is a flowchart of correlation analysis processing.

FIG. 3 is a block diagram showing a first embodiment of the present invention.

FIG. 4 is a block diagram showing a second embodiment of the present invention.

FIG. 5 is an explanatory diagram showing an example of a process waveform.

[Explanation of symbols]

1 Process input/output device 2 Data collection mechanism 3 Man-machine mechanism 4 Process control processing mechanism 5 Plant database 6 Historical database 7 Process analysis mechanism 8 Target plants 9 Input/output device 10 Data access mechanism 11 Patrol monitoring mechanism 12 Inference mechanism 13 Knowledge database

Claims (7)

[Claims] Claim 1: Capturing a measured value from a sensor installed in a plant, comparing the measured value with a preset setting value, and adjusting the amount of operation of the plant so that the deviation between the measured value and the set value is small. A plant monitoring device that monitors a plant by controlling the plant includes a set value storage means for storing time series data of the set values from the past to the present, and a set value storage means for storing time series data of the plant operation variables from the past to the present. Calculate the correlation between the manipulated variable storage means to store, the time series data of the set values stored in the set value storage means, and the time series data of the manipulated variables stored in the manipulated variable storage means, and from the calculation result. A plant monitoring device characterized by comprising a determining means for determining whether a plant is normal or abnormal, and an output means for outputting the determination result. 2. Taking a measured value from a sensor installed in the plant, comparing the measured value with a preset set value, and adjusting the amount of operation of the plant so that the deviation between the measured value and the set value is small. A plant monitoring device that monitors a plant by controlling the plant includes a set value storage means for storing time series data of the set values from the past to the present, and a set value storage means for storing time series data of the plant operation variables from the past to the present. Calculate the correlation between the manipulated variable storage means to store, the time series data of the set values stored in the set value storage means, and the time series data of the manipulated variables stored in the manipulated variable storage means, and from the calculation result. A plant monitoring device comprising a prediction means for predicting that a plant will become abnormal in the future, and an output means for outputting the prediction result. 3. A flow rate control device, characterized in that the plant monitoring device according to claim 1 or 2 is installed. 4. A gas process characterized in that the plant monitoring device according to claim 1 or 2 is installed. 5. A calorific process characterized by installing the plant monitoring device according to claim 1 or 2. 6. Taking in a measured value from a sensor installed in the plant, comparing the measured value with a preset setting value, and adjusting the amount of operation of the plant so that the deviation between the measured value and the setting value is small. In a plant monitoring method for monitoring a plant by controlling, storing time-series data of the set value from the past to the present, and storing time-series data of the plant operation amount from the past to the present, A plant monitoring method characterized by calculating a correlation between time-series data of stored set values and time-series data of manipulated variables, and determining whether the plant is normal or abnormal based on the calculation result. 7. Taking in a measured value from a sensor installed in the plant, comparing the measured value with a preset setting value, and adjusting the amount of operation of the plant so that the deviation between the measured value and the setting value is small. In a plant monitoring method for monitoring a plant by controlling, storing time-series data of the set value from the past to the present, and storing time-series data of the plant operation amount from the past to the present, A plant monitoring method comprising calculating a correlation between stored time-series data of set values and time-series data of manipulated variables, and predicting future abnormalities in the plant based on the calculation results.