JP2014178853A - Control parameter adjustment method, control parameter adjustment method and control parameter setting device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a control parameter adjustment method capable of adjusting a PID control parameter in a process system by a few persons who have a little knowledge on a plant.SOLUTION: A plant having plural PID control loops and a control parameter adjustment method used in a support server comprises the following steps: at the plant, extracting control inputs, measurement values, target values and disturbances with respect to plural PID control loops; and transmitting time-series data of the extracted control inputs, measurement values, target values and disturbances for predetermined period of time to a support server via a telecommunication network. The support server performs control performance diagnosis on each of the PID control loops based on the time-series data to determine a PID control loop which has a poor control performance and identifies the control parameter of the PID control loop while referring the time-series data. At the plant, the PID control loop is updated based on the identified control parameter.

Description

  Embodiments of the present invention include a control parameter adjustment system that adjusts parameters of a PID control loop for maintaining physical values such as flow rate, pressure, water quality, air volume, temperature, and liquid level of a process system at a constant value. The present invention relates to a control parameter setting device and a control parameter adjustment method used.

  In operation management of water treatment / water operation processes such as sewage treatment processes, sludge digestion processes, water purification processes and water supply / distribution processes, petrochemical processes, and semiconductor manufacturing processes, the specified target performance of the process is achieved. Therefore, there is a need for operations that avoid risks and lead to energy and cost savings. These processes are operated by a so-called upper system of a planning system or a support system using mathematical methods such as a planning server and a support server, and a lower system called a local system. In the lower system, physical quantities (flow rate, pressure, water quality, air volume, temperature, liquid level, etc.) can be obtained by automatic means according to instructions from the higher system or by manual means based on human judgment with reference to information from the higher system. ) Target value (referred to as SV: Set Variable). Then, an operation amount (referred to as MV: Manipulate Variable) such as a valve and a pump is feedback controlled so that a measured value (referred to as PV: Process Variable) of the target physical quantity follows the set target value.

  Conventionally, functions for realizing energy saving / cost saving and risk avoidance are often handled mainly as higher-level systems, and in some cases, should be shared at a higher management / management level than operation management. For this reason, the feedback control positioned at the lowest position tends to be considered as a main mission to operate in accordance with an instruction (= target value) from the upper level.

  In recent years, it has been recognized and reconfirmed that the performance of feedback control positioned at the lowest level plays a very important role in energy saving, cost saving and risk avoidance. In particular, much interest has been gathered in performance monitoring and adjustment of PID control, which accounts for 90% or more of feedback control used in the process industry field.

  The relationship between the control performance of PID control (feedback control) and energy / cost saving and risk avoidance will be described below. The physical quantity can be anything such as pressure and temperature, but here the dissolved oxygen concentration in the reaction tank (hereinafter referred to as DO concentration) in the sewage treatment process is the measured value, and the air supply amount (aeration amount) is the manipulated variable. Assume that the DO concentration is 1 mg / L as a target value. In the sewage treatment process, air supply called aeration is performed in order to decompose organic matter in the sewage using the activity (metabolism) of microorganisms. The electric power required for aeration is 40% or more of the electric power used in the daily operation of the sewage treatment plant. Therefore, it is desired to reduce the air supply amount as much as possible, but if it is too small, the air supply necessary for the activity of microorganisms cannot be performed, and as a result, the quality of discharged water by sewage treatment deteriorates.

  When performing constant DO concentration control in a reaction tank by PID control, when PID control is not operating sufficiently well, there are actually many places where 0 mg / L is set for 1 mg / L set as the target value. It comes to exist. Since the DO concentration of 0 mg / L means that the microorganisms cannot act, there is a concern that the quality of the discharged water will deteriorate as a result. On the other hand, the location where DO concentration exceeds 2 mg / L is also recognized. If the DO concentration of a certain level or more can be maintained, the discharged water quality can be sufficiently maintained. Therefore, a DO concentration of 2 mg / L or more is unnecessary in a normal sewage treatment process. For this reason, maintaining a high DO concentration wastes the amount of aeration air and the accompanying electric power. Thus, if the PID control is not sufficiently adjusted, the risk of deteriorating discharged water quality and the increase in cost of wasting power are caused.

  On the other hand, when the DO concentration constant control by PID control is working well, the DO concentration as a measured value is about 0.8 mg / L to 1.2 mg / L with respect to the DO concentration of 1 mg / L set as the target value. Fluctuated. Thereby, both the risk of deterioration of discharged water quality and the increase in power cost can be suppressed. However, since the average value of the DO concentration is about 1 mg / L both when the PID control is not operating sufficiently and when it is operating, the average air volume is almost the same value in both cases. . In other words, on average, it may be received as if only the risk of deterioration of the discharged water quality was avoided. On the other hand, when the PID control is operating, if the measured value is suppressed to a fluctuation of about 0.8 mg / L to 1.2 mg / L with respect to the target value of 1 mg / L, the risk of water quality deterioration is sufficiently avoided. Therefore, the target value can be lowered. For example, even if the target value is reduced to 0.5 mg / L, the fluctuation of the measured value is suppressed to about 0.3 mg / L to 0.7 mg / L, so the DO concentration does not become 0 mg / L. It is possible to avoid the risk of deterioration of the discharged water quality. This means that by adjusting the control parameters for PID control and lowering the target value, the power cost can be reduced while avoiding the risk of deterioration of the discharged water quality.

  In the above example, the DO concentration constant control of the sewage treatment process has been described as an example, but the same can be said for temperature control and pressure control, for example. In other words, from the above example, in order to maintain the control performance while minimizing the cost of the electric power of the boiler for keeping the temperature constant and the cost of the electric power of the pump for keeping the pressure constant. Therefore, it can be seen that sufficiently adjusting the control parameters of PID control plays an extremely important role.

  As described above, adjustment of PID control plays an extremely important role in energy saving / cost saving operation and risk avoidance operation, but it is not always easy to actually adjust control parameters. For example, the adjustment of control parameters in each plant often depends on the experience and know-how of local coordinators. In a plant having complicated characteristics, a skilled engineer may identify characteristics to be controlled by some method and express the identified characteristics by a mathematical model. This mathematical model is referred to as a process model, and a control parameter is adjusted by executing a simulation in which PID control is incorporated in the process model.

  As described above, the method for identifying the PID control parameter requires experience, know-how, simulation, and the like. Therefore, it is often difficult for an inexperienced local coordinator who does not have sufficient plant knowledge to understand and disseminate the control parameter identification method. As a result, control parameters are often left unadjusted. Even if the control parameters are adjusted sufficiently, the control performance often deteriorates over time due to changes in plant characteristics over time, changes in plant operating conditions, and the like.

  From this point of view, it is natural that there is a desire to automatically adjust the PID control parameters based on the characteristics of the controlled object. Thus, self-adjusting PID control has been proposed in many non-patent and patent documents. For example, the field of control called adaptive control or learning control is a system in which adaptive control or learning control is systematized for a general feedback control system. Therefore, this control method is naturally applicable to PID control, which is a special form of feedback control. For example, it is a well-known fact that a technique called a self-tuning regulator is widely used for PID control as described in Non-Patent Document 3.

  Further, in Patent Document 1, based on the idea of the adjustment method of PID control based on Generalized Minimum Dispersion Control (GMVC) used in the self-tuning regulator, it is optimum to avoid the trouble of identifying the characteristics of the target process and building a prediction model A method for identifying the correct control parameters is disclosed. Patent Document 2 discloses a method of identifying a λ parameter included in an evaluation criterion of GMVC in consideration of a trade-off between a control response output and a control input change amount. Patent Documents 3 and 4 disclose an automatic tuning method for tuning a control element in a process control network when a controller and a workstation are coupled via a communication line in a distributed control system. Furthermore, in Patent Document 5, a plurality of model models for representing the characteristics of the controlled object are prepared, PID control parameters are adjusted based on a plurality of models created using each model, and the adjusted PID A method for selecting the optimum control parameter is disclosed.

  These methods are directed to a method of automatically adjusting the PID control parameters as much as possible, and the main purpose is to reduce various engineering elements associated with the adjustment as much as possible.

  On the other hand, in a process system that is not an embedded system, the local coordinator actually adjusts the control parameters with the ultimate responsibility for each control loop. Although veteran coordinators are expected to accelerate the demand for automation in the future, it is expected that considerable time will still be required to fully automate the final adjustment. Therefore, for the time being, a system that can adjust PID control parameters by a coordinator with a smaller number of people and not necessarily abundant knowledge of the plant, or the plant manager or the operator himself / herself can perform PID control on behalf of the coordinator. There is a great need for providing a mechanism for adjusting parameters.

Japanese Patent No. 4474555 Japanese Patent No. 4528984 JP 2000-293292 A JP 2011-103140 A JP 2009-116515 A

M. Jelali, “An overview of control performance assessment technology and industrial applications”, Control Engineering Practice, 14, pp.441-466, 2006 Toru Yamamoto, “The Role of“ Control ”in Chemical Processes-Considering Energy Saving with“ Control ””, Measurement and Control, Vol. 51, no. 10, pp. 962-967, Society of Instrument and Control Engineers, 2012 Yamamoto, Kaneda, “A Design of Self-Tuning PID Controller Based on Generalized Minimum Dispersion Control Law”, Journal of System Control Information Society, 1998, Vol. 11, No. 1, pp. 1-9 http://tech.chase-dream.com/cpm/report-CPM.pdf Van Oversche, “P Subspace Identification”, Theory- Implementation-Application, PhD Thesis, Dept. of Electrical Eng. Katholieke Universiteit, Leuven, 1995 Edited by the Society of System Control Information Authors Nobuhide Suda, “PID Control”, System Control Information Library Asakura Shoten, 1992

  As described above, in the process system, the local coordinator adjusts the control parameters of each control loop. However, the PID control parameters can be set by a small number of people who have little knowledge about the plant. An adjustable mechanism is required.

  Therefore, an object of the present invention is to provide a control parameter adjustment method, a control parameter adjustment program, a control parameter adjustment system, and a control capable of adjusting a PID control parameter by a person having a small number of people and a lack of knowledge about a plant in a process system. It is to provide a parameter setting device.

  According to the embodiment, a control parameter adjustment method used in a plant having a plurality of PID control loops and a support server connected to the plant via a communication network includes the following procedure. That is, in the plant, information about the plurality of PID control loops is acquired, and an operation amount, a measured value, a target value, and disturbance (DV: Disturbance Variable) are extracted from the acquired information, and the extracted operation amount and measurement are extracted. Time series data over a predetermined period of value, target value, and disturbance is transmitted to the support server via the communication network. The support server diagnoses the control performance for each PID control loop based on the time series data, determines a PID control loop in which the control performance does not satisfy a preset condition, and determines the time series of the determined PID control loop With reference to the data, the determined control parameter of the PID control loop is identified. In the plant, the determined control parameter of the PID control loop is updated based on the identified control parameter.

It is a block diagram which shows the function structure of the control parameter adjustment system which concerns on 1st Embodiment. It is a block diagram which shows the other function structure of the identification part shown in FIG. It is a figure which shows the coefficient of the reference model used by the control parameter identification part shown in FIG. It is a figure which shows the example of a display of the display part shown in FIG. It is a figure which shows the example of a display of the display part shown in FIG. It is a block diagram which shows the function structure of the control parameter adjustment system which concerns on 2nd Embodiment. It is a block diagram which shows the function structure of the control parameter adjustment system which concerns on 3rd Embodiment.

  Hereinafter, embodiments will be described with reference to the drawings.

(First embodiment)
FIG. 1 is a block diagram showing a functional configuration of a control parameter adjustment system 10 according to the first embodiment. A control parameter adjustment system 10 illustrated in FIG. 1 includes plants 11-1 to 11 -N, a support server 12, and an operation terminal 13. The plants 11-1 to 11-N are connected to the support server 12 via a communication network. The communication network may be any communication network, but in the present embodiment, since it is assumed to be connected to the support server 12, such as the Internet (VPN) (Virtual Private Network) or the like that considers data security to some extent It is assumed that a general-purpose line or a dedicated line is used.

  The plant 11-1 is provided with a process such as a water treatment / water operation process, a petrochemical process, or a semiconductor manufacturing process. The plant 11-1 includes at least one PID control loop 111, a measurement unit 112, an extraction unit 113, a holding unit 114, and a parameter setting unit 115.

  For example, when the plant 11-1 is a sewage treatment plant, the PID control loop 111 performs DO concentration constant control using the DO concentration as a measurement value, the aeration air amount as an operation amount, and a predetermined DO concentration as a target value. Applicable to the system. Specifically, when the plant 11-1 is a sewage treatment plant, the PID control loop 11 includes a DO concentration meter that measures DO concentration, an aeration device that performs aeration, and a PID controller. The PID controller adjusts the amount of aeration air by the aeration apparatus so that the DO concentration measured by the DO concentration meter is maintained at the DO concentration input as the target value. However, the PID control loop 111 may be configured as a cascade type PID control loop in which the aeration air amount in the aeration apparatus is a target value and the valve opening degree of a valve that supplies water to the reaction tank is an operation amount.

  For example, when the plant 11-1 is a water purification plant, the PID control loop 111 corresponds to a system that performs chemical injection control of a flocculant and hypochlorous acid. For example, when the plant 11-1 is a water supply / distribution process, the PID control loop 111 corresponds to a system that performs pressure control and flow rate control of the water supply / distribution pump. For example, when the plant 11-1 is a seawater desalination process, the PID control loop 111 corresponds to a system that performs pressure control of a high-pressure pump used in the membrane treatment process. Further, when the plant 11-1 is an incineration process, the PID control loop 111 corresponds to a system that performs temperature control. Note that many of the PID control loops 111 exemplified here are configured by cascade-type PID control with the device operation amount regarding the valve opening or the rotation speed of the pump as the lowest layer.

  The measurement unit 112 acquires input information to each PID control loop 111 and output information from each PID control loop 111 at a preset cycle. When the plant 11-1 is a sewage treatment plant, the acquired information includes DO concentration, aeration air volume, target value for DO concentration, inflow sewage amount, sewage load (inflow sewage amount × inflow sewage water quality concentration), and the like. It is.

  The extraction unit 113 is composed of, for example, SCADA (Supervisory Control And Data Acquisition). The extraction unit 113 receives all measurement values acquired by the measurement unit 112. For example, the extraction unit 113 extracts a set of the operation amount, the measurement value, the target value, and the disturbance from the received measurement value by one of the following two methods. As described above, when the plant 11-1 is a sewage treatment plant, the aeration air volume corresponds to the manipulated variable, the DO concentration corresponds to the measured value, the set predetermined DO concentration corresponds to the target value, and the inflow The amount of sewage and sewage load correspond to disturbance. In this embodiment, the disturbance means a disturbance variable. When the correspondence relationship between the operation amount, the measurement value, the target value, and the disturbance is defined in advance, the extraction unit 113 determines the operation amount, the measurement value, the target from the received measurement value based on the correspondence relationship defined in advance. Extract a set of values and disturbances.

  In the first method of extracting a set of manipulated variables, measured values, target values, and disturbances from the received measured values, the extraction unit 113 searches for identification numbers of all PID control loops 111 belonging to the plant 11-1. Then, the corresponding measurement value and operation amount are extracted from the input / output information corresponding to the identification number of each PID control loop. Then, the extraction unit 113 extracts a target value corresponding to the extracted measurement value. According to this method, since the PID control loop 111 incorporated in the plant 11-1 is fully searched, there is an advantage that a set of all manipulated variables, measured values, and target values can be extracted.

  In another method, the extraction unit 113 acquires a measurement value corresponding to a target value assigned in advance from the received measurement value. The extraction unit 113 searches for an identification number of a PID control loop that uses the acquired target value and measurement value as input and output, respectively. The extraction unit 113 extracts the operation amount corresponding to the PID control loop to which the searched identification number is assigned from the received measurement value. According to this method, a PID control loop without target value information cannot be found, but it is possible to comprehensively find PID control loops that are actually used with target values set. That is, even if the number of PID control loops and the objects are not known in advance in the plants 11-1 to 11 -N, it is possible to automatically extract the PID control loop data set.

  In addition, when there are a lot of analog data other than the operation amount, the measurement value, and the target value in the received measurement value, the extraction unit 113 estimates the disturbance by analyzing the correlation between the operation amount, the measurement value, and the analog data. To do. For example, the extraction unit 113 performs principal component analysis or the like on the target operation amount, measurement value, and all analog data, and similarly to the principal component (loading) in which the operation amount and measurement value are strongly effective. Adopt analog data that works well for the disturbance.

  The holding unit 114 temporarily holds time-series data over a predetermined period of the PID control data set regarding the operation amount, measurement value, target value, and disturbance extracted by the extraction unit 113. The holding unit 114 transmits the time-series data to be held to the support server 12 via the communication network in response to a preset period or a request from the support server 12.

  The parameter setting unit 115 sets the control parameter set in the PID controllers of the plants 11-1 to 11 -N as the control parameter transmitted from the operation terminal 13 in accordance with the update instruction transmitted from the operation terminal 13. cure.

  The plants 11-2 to 11-N are similar to the plant 11-1 and have the same configuration as the plant 11-1.

  The support server 12 illustrated in FIG. 1 is connected to the plants 11-1 to 11-N via a communication network. The support server 12 includes, for example, a CPU (Central Processing Unit) and a storage area for programs and data for the CPU to execute processing such as a ROM (Read Only Memory) and a RAM (Random Access Memory). The support server 12 includes a holding unit 121, a diagnosis unit 122, and an identification unit 123 by causing a CPU to execute a control parameter adjustment program.

The holding unit 121 receives M sets of time-series data supplied from the plants 11-1 to 11-N via the communication network. Here, the M group is a time series data collected from a plurality (Li (i = 1, 2,..., N)) from the plants 11-1 to 11-N.

Is required. The holding unit 121 holds the received time series data. The holding unit 121 outputs the time series data to be held to the diagnosis unit 122 according to a preset period or an externally applied trigger.

  The diagnosis unit 122 performs control performance diagnosis on M sets of time-series data. As a control performance diagnosis method, a diagnosis using an index such as a minimum variance (MV) index or a Harris index bearing the name of the inventor is well known. It is also possible to make a diagnosis using a generalized minimum variance (GMV) index that generalizes the minimum variance index and uses both the measured value and the manipulated value. The outline of the diagnostic method using the minimum variance index is shown below. For example, Non-Patent Document 1 and Non-Patent Document 4 describe specific methods for diagnosis using the minimum variance index.

In the constant value control by feedback control having an integrator such as PID control, the target value target value and the measured value are controlled as a special case of the “internal model principle” that there is no coincidence over time. It is a well-known fact that is widely known in the technical community. However, in actual process system control, there is always an error between the measured value and the target value. This error is considered to be caused by disturbance. The minimum dispersion index η (τ) defines the control performance by the ratio of the minimum value (minimum dispersion) σ _mv ² (τ) of the error variance to the actual variance σ _y ² . The specific formula is

It is expressed. In equation (2), the minimum variance depends on the dead time τ. The dead time is a delay time until the influence of disturbance appears in the output. Even if any control is performed during the dead time after the disturbance is applied, the influence cannot be suppressed. Based on this fact, the minimum dispersion is obtained as the minimum dispersion that cannot further reduce the dispersion. However, since the minimum variance depends on the dead time τ, the dead time τ is estimated in advance, for example, by calculating the correlation between the manipulated variable and the measured value while shifting the time and estimating the dead time with the maximum correlation. It is necessary to keep it. The minimum dispersion index η (τ) takes a value from 0 to 1, and the closer to 1, the better the control performance is diagnosed. Therefore, the diagnosis unit 122 diagnoses whether the performance is good by determining whether η (τ) exceeds a predetermined threshold.

The generalized minimum variance index is an index that considers not only the variance σ _y ² of the measurement value but also the variance of the manipulated variable (difference thereof), but the basic idea is the same as the minimum variance index.

  As described above, it is possible to diagnose the control performance by a known diagnostic method using the minimum variance index and the generalized minimum variance index, but the minimum variance index and the generalized minimum variance index have also been pointed out. It may not be practical. In such a case, the diagnosis unit 122 can also take the following method.

The simplest method is to diagnose the control performance by comparing the variance at the time of initial adjustment with the current variance when the PID control parameters have already been adjusted once and the control performance has deteriorated due to changes over time, etc. It is a method to do. A specific equation is obtained from the ratio between the variance σ _bench ² of the control error between the measured value and the target value at the initial adjustment and the actual variance σ _y ² .

It is expressed. _ηbench takes a value from 0 to 1, and a value closer to 1 is diagnosed as having better control performance. Therefore, the diagnosis unit 122 diagnoses whether the performance is good or not by determining whether _ηbench exceeds a predetermined threshold value. Accordingly, the diagnosis unit 122 can evaluate the performance based on the control response at the time of initial adjustment when evaluating the aging of the control parameter when the control parameter of the plant has already been adjusted. The control performance of the PID control loop can be diagnosed while solving the problems of the minimum dispersion index and the generalized minimum dispersion index.

By the way, there are many cases where the PID control parameters are not adjusted sufficiently well in the first place. In such a case, a practical method for diagnosing control performance is to compare the dispersion of PID control loops (hereinafter referred to as the same type of PID control loops) having the same operation amount, measurement value, and target value. The method of doing is mentioned. For example, in the case of a water treatment process, DO concentration constant control is often performed in a plurality of series called water treatment series. Even if a single water treatment sequence is executed and the DO concentration constant control is executed, if information of a plurality of plants is collected in the support server 12, a plurality of PID control loops of the same type usually exist. In such a case, the diagnosis unit 122 compares the error variance between the measured value and the target value of the same type of PID control loop, and is closest to the average value of the error variance in the same type of PID control loop, for example. The error variance is the benchmark target. The diagnosis unit 122 diagnoses the quality of the control performance from the ratio of the control error variance _σbench ² to the benchmark target and the actual variance σ _y ² . Thereby, when there is a control loop of a plurality of plants having the same control loop and / or similar plant configuration, the diagnosis unit 122 can perform the performance evaluation by relatively comparing the control responses. Therefore, it is possible to diagnose the control performance of the PID control loop while solving the problems of the minimum variance index and the generalized minimum variance index. Note that standard deviation may be used instead of variance.

  The diagnosis unit 122 can also perform the following diagnosis method as an improved version of the above-described control performance diagnosis.

  A water quality sensor or the like included in the measurement unit 112 in a water treatment plant often includes a considerable outlier (abnormal value) in the measurement value of the sensor due to adhesion of contaminants. In some cases, the presence of such outliers hinders practical use of feedback control for water quality control. For this reason, applying feedback control after filtering outliers is fundamental in performing feedback control, and should be performed before driving feedback control. However, no matter how the outlier is filtered, it is difficult to completely remove the outlier. Therefore, the control response is often disturbed, and the control error between the target value and the measured value may increase. In such a case, the diagnosis unit 122 may make a diagnosis of control performance using a value that is robustly estimated instead of the variance shown in Expression (2) and Expression (3). In the field of robust statistics, various methods have been developed to robustly estimate variance. Here are some of the simplest and most commonly used techniques:

  First, instead of the mean-variance pair, a median (Median) -median absolute deviation (MAD) is used as a statistical measure. When it is assumed that the target data (in this case, the control error between the target value and the measured value) follows a normal distribution, MAD and variance can be converted, and therefore MAD can be used instead of variance. Note that MAD has the property that even if the data within 50% is replaced with infinity (that is, it is assumed that there are over / underly abnormal values within 50%), the value remains finite. It is known to be a robust statistic. As a result, the diagnosis unit 122 can diagnose the control performance by replacing the dispersion represented by the equations (2) and (3) with the MAD.

  Also, as a practical method that is not as robust as median absolute value deviation but more robust than variance, pruning that sorts the data in ascending or descending order and trims the upper and lower limits to calculate the mean and variance (Trim) calculation. Although the trim calculation tends to underestimate the variance, it does not appear to have a significant impact on the relative variance ratio assessment. Therefore, the diagnosis unit 122 can diagnose the control performance by replacing the variances of the expressions (2) and (3) with the trimmed values.

  In addition, it is possible to evaluate the dispersion using a technique called bootstrap and subsampling. Various specific methods have been proposed, but the simple and effective method is to extract two data from all data, calculate the square of the difference over all patterns, and then calculate the median Is a way to take. This is a method that extends the estimation method called the Hodges-Lehman statistic, which is a robust calculation of the average, to a variance. However, if the target data does not deviate significantly from the normal distribution, the estimate is a good variance. . As a result, the diagnosis unit 122 can accurately evaluate the control performance even when abnormal data such as outliers is mixed in the time-series data.

  Further, as a further improved method for diagnosing the control performance described above, the diagnosis unit 122 includes an absolute index that is theoretically calculated from one PID control loop, such as a minimum variance index and a generalized minimum variance index. It is also possible to evaluate the performance in combination with a relative index obtained from a relative comparison of a plurality of PID control loops.

  When there are a plurality of PID control loops of the same type, it is not a bad value when the control performance of a certain PID control loop is evaluated using the minimum variance index, but it is a bad value when evaluated relative to other PID control loops. May be. For example, when the DO concentration constant control is performed in a plurality of water treatment series of the water treatment process, the dispersion is relatively large, and the PID control loop diagnoses that the performance using the relative index is poor. However, the diagnosis using the minimum variance index does not necessarily diagnose that the performance is bad. This is because the minimum variance index uses the variance based on the uncontrollable disturbance as the minimum variance and the benchmark target.

  However, the PID control loop having a relatively large dispersion often has a problem in the performance of the actuator and / or the sensor installation position. For example, when the DO concentration meter is installed near the aeration device that is an actuator with DO concentration constant control, the air supplied from the aeration device directly hits the DO concentration meter, so that the DO concentration can change greatly. On the other hand, the DO concentration actually dissolved in water does not change so rapidly. This situation can be improved by reviewing the installation position of the DO densitometer. Thus, when there are PID control loops having different control performance in the same type of PID control loop, it is often necessary to review the installation positions of sensors and / or actuators installed in the PID control loop. .

  Therefore, the diagnosis unit 122 diagnoses the control performance using an absolute index such as the minimum variance index shown in Expression (2), and after determining that the control performance is good based on this diagnosis, the diagnosis section 122 shows in Expression (3). Diagnose control performance using relative indicators. When it is determined that the control performance is poor by the diagnosis using the relative index, the diagnosis unit 122 prompts the review of the installation position of the sensor and / or the actuator installed in the PID control loop 111 instead of adjusting the control parameter. Then, an instruction is displayed on the operation terminal 13. Accordingly, the diagnosis unit 122 can diagnose whether or not the control parameter adjustment is necessary, and at the same time, whether or not the device configuration such as the actuator and / or the sensor needs to be reviewed.

  The diagnosis unit 122 grasps time series data with poor control performance by performing the above diagnosis on the M sets of time series data output from the holding unit 121. The diagnosis unit 122 extracts a PID control loop in which time series data with poor control performance is measured as a PID control loop that requires adjustment of control parameters. The diagnosis unit 122 outputs information about the PID control loop that requires adjustment of the control parameter to the identification unit 123.

  Based on the information notified from the diagnosis unit 122, the identification unit 123 reads time-series data regarding a PID control loop that requires adjustment of control parameters from the holding unit 121. The identification unit 123 identifies the control parameter based on the read time series data. The identification unit 123 uses a closed loop identification method that does not require an identification experiment. The closed loop identification method is roughly classified into three types, that is, a direct method, an indirect method, and an input / output coupling method, and any method may be used. As a theoretically refined method, for example, a method such as a subspace method described in Non-Patent Document 5 can be used.

  Hereinafter, a case where the identification unit 123 identifies the control parameter using the indirect method and a case where the control parameter is identified using the direct method will be described.

  First, when the identification unit 123 identifies a control parameter using the indirect method, the identification unit 123 includes a process model identification unit 1231 and a control parameter identification unit 1232. FIG. 1 shows a configuration when the identification unit 123 employs the indirect method.

For example, the process model identification unit 1231 identifies the process model by approximation with the first order delay + dead time model shown in Expression (4).

  The process model identification unit 1231 uses the closed loop identification method for the time-series data read from the holding unit 121, so that the process parameters, K (process gain), T ( Three parameters of time constant) and L (dead time) are identified. The process model identification unit 1231 uses the closed-loop identification method, so that even if only time-series data in a state where PID control is applied to the PID control loop exists, the plant 11-1 to 11-11 are not performed again. It is possible to identify process parameters only from time series data collected from -N.

  By the way, the closed loop identification method is a useful method that can identify parameters of a process model without performing an identification experiment again. However, when the identifiability of process parameters (the property that parameters such as K, T, and L are uniquely determined) is deteriorated, the process parameters cannot always be correctly identified. In order to cope with such a case, a plurality of methods shown below can be considered.

  The first method is a method of selectively using a period in which the operation amount is changing among the time-series data read from the holding unit 121.

  When realizing a method of selectively using the period during which the operation amount is changed, the process model identification unit 1231 provides a threshold value for the change rate of the operation amount, for example, out of the time-series data read from the holding unit 121 Then, data for a period in which the operation amount is changing more than a threshold value is extracted. Then, the process model identification unit 1231 identifies process parameters using the extracted data.

  Moreover, the method of selectively using the period during which the operation amount is changed can be realized as follows. That is, the process model identification unit 1231 searches the time series data read from the holding unit 121 for a period in which the operation amount is changed in steps, and extracts a step response in the vicinity of the found period. At this time, for example, the process model identification unit 1231 extracts a period that changes stepwise using a wavelet transform such as a Harr wavelet that easily detects a step change. Then, the process model identification unit 1231 identifies process parameters using the extracted data.

  Moreover, the method of selectively using the period during which the operation amount is changed can be realized as follows. That is, the process model identification unit 1231 repeats dividing the time-series data read from the holding unit 121 into an appropriate length and merging the divided data (Split and Merge). The process model identification unit 1231 confirms the identifiability of the merged time series data by applying a direct method to the operation amount and the measurement value in the time series data. Then, the process model identification unit 1231 identifies process parameters using time series data having the highest identifiability among the merged time series data.

  As a result, the process model identification unit 1231 can identify process parameters with high accuracy by extracting time-series data that facilitates process parameter identification from time-series data with low identifiability of process parameters. It becomes.

  In addition, the second method for deterioration of identifiability of process parameters is a method for forcibly improving identifiability.

  When realizing the second method, the process model identification unit 1231 first confirms the identifiability when using the time-series data held in the holding unit 121 and determines the degree of deterioration of identifiability. . This is determined, for example, by providing a threshold for the condition number represented by the ratio between the maximum eigenvalue and the minimum eigenvalue of the matrix for determining identifiability. If it is determined that the identifiability is bad, the process model identification unit 1231 instructs the operation terminal 13 to take an action for improving the identifiability. That is, the process model identification unit 1231 temporarily opens the PID control loop (switches from the automatic mode to the manual mode) for a plant in which an identification experiment is allowed, and inputs a step input to the operation amount, or The operation terminal 13 is instructed to input the M-sequence signal. The process model identification unit 1231 obtains the time series data about the control response to this action, and obtains the process parameters of the process model by open loop identification.

  On the other hand, it is considered that there are not many cases where the above identification experiment is allowed. Therefore, the process model identification unit 1231 can take the following processing. That is, the process model identification unit 1231 instructs the operation terminal 13 to weaken the PID control of the PID control loop once and to deteriorate the control performance. For example, the process model identification unit 1231 instructs the operation terminal 13 to degrade the control performance by decreasing the proportional gain and increasing the integration time. As a result, the control becomes less effective, and the fluctuation between the manipulated variable and the measured value becomes large, and it can be expected that the identifiability is improved. The process model identification unit 1231 collects control responses with weak control for a certain period, confirms the identifiability again, and repeats this operation until the identifiability is sufficient. Then, the process model identification unit 1231 identifies the process parameter using the closed loop identification method when the identifiability is sufficient.

  As a result, the process model identification unit 1231 can identify process parameters for time-series data for which the parameter identifiability is forcibly improved.

  As shown in FIG. 2, the identification unit 123 should shift the process model identification unit 1231 to the control parameter identification process from the value of the identified process parameter, or control to control the constituent devices in the PID control loop. You may make it provide the determination part 1233 which judges whether a container should be reviewed. The determination unit 1233 performs the following procedure.

  The process model identification unit 1231 obtains process parameters K, L, and T. The L / T value obtained from these values is often used as an index for evaluating the difficulty of stabilization by PI control. The larger the process gain K is, the more difficult it is to control, but since it can be compensated by reducing the proportional gain Kp, which is one of the PI control parameters, the process gain K has little to do with the essential difficulty. L / T means that the larger the absolute value, the longer the dead time compared to the time constant. That is, the larger the absolute value of L / T, the more difficult the control. Although it is not so many in the process system, in an extreme case, if the absolute value of L / T exceeds 1 in an unstable process (T value is negative), the plant cannot be stabilized by PID control. Although there are few unstable plants, control is difficult when the value of L / T is large, and therefore, a control method that more positively considers dead time may be preferable. Therefore, the determination unit 1233 provides, for example, a threshold value “5” to the absolute value of L / T, and determines whether or not the value of L / T exceeds the threshold value. When exceeding, the determination part 1233 announces to the operating terminal 13 to review PID control, without shifting to control parameter identification processing. Here, reviewing the PID control refers to, for example, a dead time compensation type Smith compensation controller or a Dahlin controller known to be robust against dead time, instead of the current controller. It means adopting a controller.

  As a result, if PID control alone is not enough to improve control performance, improving the controller itself of the PID control loop or changing the controller will increase the effect of energy saving, cost saving and risk reduction. Is possible.

  The control parameter identification unit 1232 identifies the control parameter using the process parameter identified by the process model identification unit 1231. For this identification, any identification method using process parameters of the process model can be used.

  For example, Non-Patent Document 6 describes a method for adjusting various control parameters with respect to the process model represented by Expression (4). In the present embodiment, any one of those methods may be used. Among these, the method considered particularly promising is the partial model matching method (Kitamori method) proposed by Kitamori or the IMC adjustment method (internal model adjustment method). However, the Kitamori method is not necessarily an adjustment method for Equation (4). Therefore, in the following, a specific calculation formula of the Kitamori method when the PI control excluding the D component is applied to the formula (4) is shown.

First, as step 1, in order to determine the values of the proportional gain Kp and the integral constant TI of the PI controller, the process model of Expression (4) is approximately expressed in the form of Expression (5). Here, Equation (5) is called a denominator series expression having a polynomial of the Laplace operator S only in the denominator of the transfer function.

In equation (5),

The following relational expression was used.

Subsequently, as step 2, a reference model is defined. This is because the PI control parameters are adjusted so that the denominator series expression of Equation (5) matches the step response of the reference model. The reference model is

It is expressed. The values of r ₀ to r _{3 in} Equation (7) are No. 1 shown in FIG. 1-No. Any one of 5 values is selected.

Subsequently, as step 3, the proportional gain of the PI control is set so that the feedback control system that performs the PI control on the process model shown in Expression (5) matches the response waveform of the reference model shown in Expression (7). Kp, an integral constant TI, and an adjustable parameter t that defines the response speed of the reference model are determined. When the proportional gain Kp, the integration constant TI, and the adjustable parameter t are determined, the following relational expression is obtained.

  The control parameter identification unit 1232 identifies the control parameters of the PID control loop by performing the calculations of the equations (4) to (10) shown above.

  For comparison, an IMC adjustment method targeting Equation (4) is also shown below.

In the IMC adjustment method, the proportional gain Kp, the integral constant TI, and the adjustable parameter t are obtained from the following equations.

  The control parameter identification unit 1232 identifies the control parameters of the PID control loop by performing the calculations of the equations (11) to (13) shown above.

  Next, a case where the identification unit 123 identifies the control parameter using the direct method will be described. The direct method is employed when it is difficult to identify process parameters. In this embodiment, since the purpose is to adjust the control parameters of PID control, it is not always necessary to identify the process model. Therefore, a control parameter is directly identified from the closed loop data by using a technique called IFT (Iterative Feedback Tuning) or VRFT (Virtual Reference Feedback Tuning). If this method is used, the process model identification part 1231 becomes unnecessary. This makes it possible to adjust control parameters without performing a control response test even when only time-series data in a state where PID control is applied to the PID control loop exists. In addition, when it is difficult to identify the process parameter from the control response data from the PID control loop, it becomes possible to directly identify the control parameter with higher accuracy than to obtain the control parameter from the process parameter. . However, since it may be necessary for the adjuster of the control parameter to know to some extent the values of the process gain K, the time constant T, and the dead time L that represent the response on the process side, an indirect method that can calculate the process parameter. Is more preferable.

  When identifying the control parameter by the indirect method, the identifying unit 123 transmits the process parameter identified by the process model identifying unit 1231 to the operation terminal 13 in addition to the identified control parameter. When the control parameter is identified by the direct method, the identification unit 123 transmits the identified control parameter to the operation terminal 13 via the communication network.

  The operation terminal 13 is a notebook personal computer, a tablet personal computer, a smartphone, a dedicated terminal, or the like, and there is no limitation on the embodiment. The operation terminal 13 is carried when a PID control parameter adjuster and / or a plant manager / operator is physically away from the actual plants 11-1 to 11 -N and the support server 12. The operation terminal 13 includes a display unit 131, an input unit 132, and an update processing unit 133.

  The display unit 131 receives information transmitted from the support server 12 and displays the received information. Specific display examples on the display unit 131 are shown in FIGS.

  As shown in FIG. 4, the display unit 131 displays the identification result (nominal value) of the process parameter identified by the process model identification unit 1231. At this time, it is preferable that the display unit 131 displays the maximum value and the minimum value of the process parameter together with the identification result. In order to display the maximum value and the minimum value, the process model identification unit 1231 repeatedly identifies process parameters for a plurality of time-series data in advance, and estimates the width of the identified process parameters.

  Further, the display unit 131 displays the identification result (optimum value) of the control parameter identified by the control parameter identification unit 1232. At this time, it is preferable to compare and display the current value of the control parameter together with the identification result.

  Further, as shown in the lower part of FIG. 5, the display unit 131 displays the stable region of the PI control parameter, the current control parameter value, and the identified control parameter value. The left diagram in the lower part of FIG. 5 is a diagram showing the stable region when the PI control is applied in relation to the proportional gain Kp and the integration time TI, and the white region corresponds to the stable region. Note that a region surrounded by a square in the stable region shows a preferable range of parameter values in the stable region. Further, the right diagram in the lower part of FIG. 5 is a diagram showing the stable region when the PI control is applied in relation to the proportional gain Kp and the integral gain KI, and the white region corresponds to the stable region. Further, in the left and right diagrams in the lower part of FIG. 5, the cross marks indicate the current control parameter values, and the circles indicate the identified control parameter values. By displaying in this way, it is possible to visually confirm how bad or how good the current control parameter and the identified control parameter are.

  Moreover, the display part 131 displays the step response waveform of a PID control loop as shown in the upper stage of FIG. The left figure in the upper part of FIG. 5 shows a step response waveform according to the current control parameter, and the right figure in the upper part of FIG. 5 shows a step response waveform according to the identified control parameter. In addition, the step response waveform in the upper part of FIG. By displaying in this way, it is possible to visually confirm that the control performance is improved.

  The coordinator determines whether to adopt the identified control parameter while referring to the display content displayed on the display unit 131. When it is determined that the identified control parameter is to be adopted, the coordinator inputs to the input unit 132 an update instruction for updating the control parameter in the plants 11-1 to 11-N to the identified control parameter. When the adjuster looks at the display content of the display unit 131 and determines that further fine adjustment is necessary, the adjuster changes the control parameter displayed as the optimum value in FIG. Then, the coordinator inputs an update instruction for updating the control parameters in the plants 11-1 to 11 -N to the control parameters after the change in the input unit 132.

  When an update instruction is input to the input unit 132, the update processing unit 133 transmits the identified control parameter together with the input update instruction to the parameter setting unit 115 of the plants 11-1 to 11 -N via the communication network. Send to.

  As described above, in the control parameter adjustment system according to the first embodiment, the plants 11-1 to 11-N are connected to the support server 12 via the communication network. That is, the support server 12 plays a role of a cloud server connected to a plant via a communication network. The so-called cloud mechanism is attracting attention as a framework for an inexpensive monitoring and control system that replaces the conventional monitoring and control system. In fact, an increasing number of companies are realizing monitoring services and information providing services using cloud servers. The support server 12 according to the present embodiment receives time series data of the PID control data set for the operation amount, the measured value, the target value, and the disturbance measured in the plants 11-1 to 11-N. The support server 12 identifies the control parameters of the PID control loop in the plants 11-1 to 11-N based on the received time series data. As a result, it is possible to support the adjustment of the PID control parameter using the cloud mechanism including the support server 12.

  In the first embodiment, the operation terminal 13 is carried by the PID adjuster, and the control parameter and / or the process parameter identified by the support server 12 is received. As a result, the user of the operation terminal 13 can monitor the control state of the plant and adjust the PID control parameters anytime and anywhere regardless of the location and time.

  Therefore, according to the control parameter adjustment method according to the present embodiment, the PID control parameter can be adjusted by a lesser number of people in the process system and by a person who has little knowledge about the plant. In other words, a mechanism for efficiently adjusting the PID control parameters has been established, and it is possible to reduce the adjustment manpower of the local coordinator and to make efficient and optimal adjustment of the PID control parameters by the plant manager / operator himself when there is no coordinator. Become. For this reason, the plant operation that achieves both energy saving / cost saving operation and risk reduction becomes possible.

  In the first embodiment, the case where no limit is set on the value of the operation amount after the adjustment of the control parameter has been described as an example. However, it is not limited to this. For example, when the PID control loop 111 has a cascade structure, the identification unit 123 may set an upper limit value and a lower limit value for the lower operation amount in advance when adjusting the control parameter. When the control performance in the PID control loop is poor, the reason is often due to the movement of the operation amount. Such a phenomenon is recognized when the PID control has a cascade structure. For example, when the DO concentration constant control is performed in the water treatment process, if the control parameter of the DO concentration constant control is adjusted in order to improve the control performance, the value of the air volume that is the operation amount is greatly changed compared to before adjustment. There is a case. The valve opening degree-air volume PID control is inserted in a cascade structure as the lower PID control of the air volume, which is the operation amount of the DO concentration constant control, and the valve opening greatly changes as the air volume changes greatly. . If the valve is greatly operated in the closing direction, the opening may become zero, which destabilizes the plant. On the other hand, when the valve opening suddenly operates in the opening direction, the opening operation may be delayed because the valve is opened at an opening greater than a certain value as the valve characteristic, even though the air volume does not increase. In order to avoid such a phenomenon, the control adjustment may be loosened intentionally. Therefore, if the identified control parameters are reflected as they are, the plant may become unstable. To avoid such a phenomenon, set the maximum and minimum values of the valve opening before adjustment in advance, or keep the values from the maximum to several percent and from the minimum to several percent. When the control parameter is adjusted, the identification unit 123 sets this value as the upper and lower limit values of the valve opening. This makes it possible to prevent the plant from becoming unstable due to excessive operation of devices such as valves and pumps when the control parameters are adjusted. That is, the control parameters can be adjusted while keeping the plant in a stable state.

  In the first embodiment, the support server 12 transmits control parameters and / or process parameters to the operation terminal 13, and the operator of the operation terminal 13 updates the control parameters of the plants 11-1 to 11-N. The case has been described as an example. However, it is not limited to this. For example, the identification unit 123 transmits the control parameter and / or the process parameter to the operation terminal 13 after completing the update of the control parameter for the plants 11-1 to 11-N, and the coordinator and the plant manager / operator. You may be prompted to confirm. When such a configuration is adopted, it is possible to update the control parameters almost automatically, although there is a slight risk due to delaying confirmation.

(Second Embodiment)
FIG. 6 is a block diagram illustrating a functional configuration of the control parameter adjustment system 20 according to the second embodiment.

  The control parameter adjustment system 20 includes plants 11-1 to 11 -N and a cloud server 21.

  The cloud server 21 includes a holding unit 121, a diagnosis unit 122, an identification unit 123, a display unit 211, an input unit 212, and an update processing unit 213. Note that the display unit 211, the input unit 212, and the update processing unit 213 may be provided in a location near the cloud server 21.

  The display unit 211 receives control parameters and / or process parameters identified by the identification unit 123. The display unit 211 displays the same content as the display unit 131 described in the first embodiment.

  The coordinator determines whether to adopt the identified control parameter while referring to the display content displayed on the display unit 211. When it is determined that the identified control parameter is to be adopted, the coordinator inputs, to the input unit 212, an update instruction for updating the control parameter in the plants 11-1 to 11-N to the identified control parameter. Further, when the adjuster looks at the display content of the display unit 211 and determines that further fine adjustment is necessary, the adjustment parameter is changed by the input unit 212. Then, the coordinator inputs to the input unit 212 an update instruction to update the control parameters in the plants 11-1 to 11-N to the changed control parameters.

  When an update instruction is input to the input unit 212, the update processing unit 213 transmits the identified control parameters together with the input update instruction to the plants 11-1 to 11-N via the communication network.

  As described above, in the second embodiment, the display unit 211, the input unit 212, and the update processing unit 213 are provided in a place where the cloud server 21 physically exists or in the vicinity thereof, and the PID parameter is set in the place. A small number of coordinating experts are deployed. This is a cloud service that assumes a business form that undertakes PID control adjustments, and assumes that several coordinators adjust PID control parameters for many plants and provide services for improving plant operations. It is a thing.

(Third embodiment)
FIG. 7 is a block diagram illustrating a functional configuration of the control parameter adjustment system 30 according to the third embodiment.

  The control parameter adjustment system 30 includes plants 31-1 to 31 -N and the support server 12.

  The plant 31-1 includes a PID control loop 111, a measurement unit 112, an extraction unit 113, a holding unit 114, a display unit 311, an input unit 312, an update processing unit 313, and a parameter setting unit 115.

  The plant 31-1 receives the control parameters and / or process parameters identified by the support server 12 via a communication network. The display unit 311 is, for example, a SCADA monitoring screen installed in the plant 11-1, and displays the same contents as the display unit 131 described in the first embodiment based on the received information.

  The coordinator determines whether to adopt the identified control parameter while referring to the display content displayed on the display unit 311. When it is determined that the identified control parameter is to be adopted, the coordinator inputs to the input unit 312 an update instruction for updating the control parameter in the plant 31-1 to the identified control parameter. Further, when the adjuster looks at the display content of the display unit 311 and determines that further fine adjustment is necessary, the adjustment parameter is changed by the input unit 312. Then, the coordinator inputs to the input unit 312 an update instruction for updating the control parameter in the plant 31-1 to the changed control parameter.

  When an update instruction is input to the input unit 312, the update processing unit 313 outputs the identified control parameter to the parameter setting unit 314 together with the input update instruction.

  The parameter setting unit 314 resets the control parameters set in the PID controller of the plant 11-1 according to the update instruction output from the update processing unit 313.

  As described above, in the third embodiment, the display unit 311, the input unit 312, and the update processing unit 313 are provided in the plant. As a result, the plant manager / operator placed in each plant of the plant group can update the control parameters of the PID control loop. This assumes a support service in which the plant manager / operator can adjust the PID control parameters when there is no local coordinator or local coordinator for PID control.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the scope and gist of the invention, and are also included in the invention described in the claims and the equivalents thereof.

  DESCRIPTION OF SYMBOLS 10,20,30 ... Control parameter adjustment system, 11-1 to 11-N, 31-1 to 31-N ... Plant, 111 ... PID control loop, 112 ... Measurement part, 113 ... Extraction part, 114 ... Holding part, DESCRIPTION OF SYMBOLS 115 ... Parameter setting part, 12, 21 ... Support server, 121 ... Holding part, 122 ... Diagnosis part, 123 ... Identification part, 1231 ... Process model identification part, 1232 ... Control parameter identification part, 1233 ... Determination part, 13 ... Operation Terminal, 131 ... Display unit, 132 ... Input unit, 133 ... Update processing unit, 211 ... Display unit, 212 ... Input unit, 213 ... Update processing unit, 311 ... Display unit, 312 ... Input unit, 313 ... Update processing unit, 314 ... Parameter setting section

Claims (22)

In a plant having multiple PID control loops,
Obtaining information about the plurality of PID control loops;
From the acquired information, an operation amount, a measurement value, a target value, and a disturbance are extracted, and the extracted operation amount, the measurement value, the target value, and time series data over a predetermined period of the disturbance are transmitted via a communication network. Send it to the support server,
In the support server,
Diagnosing the control performance of the plurality of PID control loops based on the time series data, determining a PID control loop in which the control performance does not satisfy a preset condition;
With reference to the time series data of the determined PID control loop, the control parameter of the determined PID control loop is identified,
In the plant,
A control parameter adjustment method for updating the determined control parameter of the PID control loop based on the identified control parameter. In the plant,
From the acquired information, extract all operation amounts, measurement values and target values associated with the plurality of PID control loops,
The control parameter adjustment method according to claim 1, wherein information having a high correlation with the operation amount and the measurement value is extracted as the disturbance from the acquired information. In the plant,
The target value is acquired in advance, and a measurement value and an operation amount corresponding to the target value are extracted from the acquired information,
The control parameter adjustment method according to claim 1, wherein information having a high correlation with the operation amount and the measurement value is extracted as the disturbance from the acquired information. In the support server,
Identifying a process model based on the determined time series data of the PID control loop, calculating a process parameter for the identified process model;
The control parameter adjustment method according to claim 1, wherein a control parameter of the determined PID control loop is identified using the calculated process parameter. The control server diagnoses the control performance by comparing the error variance of the measured value with respect to the target value in the time series data and the error variance of the measured value with respect to the target value at the time of initial adjustment of the PID control loop in the support server. Item 2. The control parameter adjustment method according to Item 1. In the support server, with respect to the error of the measurement value with respect to the target value in the time series data, the robustness method calculates the error of the measurement value with respect to the target value at the time of initial adjustment of the PID control loop. The control parameter adjustment method according to claim 1, wherein the control performance is diagnosed by comparing with a value calculated using a statistical technique. In the support server,
For a plurality of PID control loops of the same type, calculate the error variance of the measured value with respect to the target value,
The error variance closest to the average value of the calculated error variances is the benchmark target,
The control parameter adjustment method according to claim 1, wherein the control performance is diagnosed by comparing an error variance of a measured value with respect to a target value in the time series data and an error variance targeted for the benchmark. In the support server, for a plurality of PID control loops of the same type, a statistical value using a robust statistical method is calculated for an error in the measured value with respect to the target value,
The statistical value closest to the average value of the plurality of calculated statistical values is set as a benchmark target,
The control parameter adjustment method according to claim 1, wherein the control performance is diagnosed by comparing a statistical value of a measured value with respect to a target value in the time series data and a statistical value as the benchmark target. In the support server,
Diagnosing the control performance of the PID control loop using an absolute index calculated from the time series data,
If it is determined that the control performance satisfies a preset condition in the diagnosis using the absolute index, the control performance of the PID control loop is diagnosed using a relative index for relatively evaluating the PID control loop;
When determining that the control performance does not satisfy a preset condition in the diagnosis using the relative index,
The control parameter adjustment method according to claim 1, wherein a resetting of a configuration installed in the PID control loop is instructed. In the support server,
The control parameter adjustment method according to claim 4, wherein a process parameter for the identified process model is calculated using a closed loop identification method. In the support server,
The control parameter adjustment method according to claim 4, wherein the process parameter is identified based on time-series data of a period in which a change in an operation amount satisfies a preset condition among the time-series data. In the support server,
The control parameter adjustment method according to claim 4, wherein the process parameter is identified based on time-series data after forcibly improving the identifiability on the PID control loop. In the support server,
Referring to the identified process parameters to determine whether it is difficult to stabilize the PID control loop;
The control parameter adjustment method according to claim 4, wherein when stabilization is difficult, a review of PID control in the PID control loop is instructed. In the support server,
The control parameter adjustment method according to claim 4, wherein the control parameter is identified from the calculated process parameter using a partial model matching method or an IMC adjustment method. In the support server,
The control parameter adjusting method according to claim 1, wherein when the PID control loop has a cascade structure, an upper limit value and a lower limit value are set for a lower operation amount when adjusting the control parameter.   The control parameter adjustment method according to claim 1, wherein the control parameter identification result, a control response when the control parameter is used, and a position of the control parameter in a control stable region are displayed.   The control parameter adjustment method according to claim 4, wherein the identification result of the control parameter and the process parameter, a control response when the control parameter is used, and a position of the control parameter in a control stable region are displayed. The identified control parameter is displayed on an operation terminal located at a location away from the support server and the plant,
The control parameter adjustment method according to claim 1, wherein an update instruction based on the displayed identification result is received from the operation terminal. The identified control parameter is displayed on a display device installed in or near the support server,
The control parameter adjustment method according to claim 1, wherein an update instruction based on the identification result to be displayed is received from an input unit connected to the display device. The identified control parameter is displayed on a display device installed in the plant,
The control parameter adjustment method according to claim 1, wherein an update instruction based on the identification result to be displayed is received from an input unit connected to the display device.   The control parameter adjustment method according to claim 1, wherein the support server is a cloud server. In a control parameter adjustment system comprising a plant having a plurality of PID control loops and a support server connected to the plant via a communication network,
The plant is
A measurement unit for obtaining information about the plurality of PID control loops;
A manipulated variable, measured value, target value, and disturbance are extracted from the acquired information, a holding unit that holds time series data over a predetermined period of the extracted manipulated variable, measured value, target value, and disturbance, and the PID control loop A parameter setting unit for updating the control parameter based on the control parameter identified by the support server;
The support server
Receiving the time series data held in the plant via the communication network, holding unit that holds the received time series server;
A diagnosis unit that diagnoses control performance for each of the PID control loops based on time-series data output from the holding unit, and determines a PID control loop in which the control performance does not satisfy a preset condition;
A control parameter adjustment system comprising: an identification unit that identifies the control parameter of the determined PID control loop with reference to the determined time series data of the PID control loop.

Images