A method and system for self-adjusting parameters of a PID control system

Abstract

This application discloses a method and system for parameter self-adjustment of a PID control system, applicable to the field of industrial control technology. In this application, first model parameters are obtained by optimizing at least one candidate model parameter. Based on the mapping relationship between the first model parameters and the target parameter values ​​of the PID parameters, the target parameter values ​​of the PID parameters are calculated. Using fuzzy theory, the rate of change of the PID parameters is calculated. The difference between the current parameter value and the target parameter value, along with the magnitude of the rate of change, is determined, and the PID parameters are then updated. In this way, the parameters of the PID control system can be adjusted in a timely manner according to changes in the dynamic characteristics of the industrial production process. Therefore, the stability of the closed-loop system of the PID control system can be improved.

Description

Technical Field

[0001] This application relates to the field of industrial control technology, and in particular to a method and system for parameter self-adjustment of a PID control system. Background Technology

[0002] Stable control of industrial production processes is fundamental to industrial process control. Parameters in industrial production processes can deviate from target values ​​due to various factors, necessitating a PID control system to adjust these parameters in a closed-loop manner to the target value. Traditional PID control systems, whose parameters are set but not changed, can generally meet the control requirements of industrial production processes.

[0003] In general, the parameters of the controlled object in a traditional PID control system are fixed. However, in actual production processes, the dynamic characteristics of industrial production equipment are constantly changing due to various factors such as environmental disturbances, equipment load, and catalyst deactivation. Such a PID control system cannot adjust its parameters according to these dynamic characteristics, leading to instability in the closed-loop system. Summary of the Invention

[0004] To address the aforementioned issues, this application provides a method and system for parameter self-adjustment of a PID control system, which can improve the stability of the closed-loop system of the PID control system.

[0005] The embodiments of this application disclose the following technical solutions:

[0006] In a first aspect, this application provides a method for parameter self-tuning of a PID control system, comprising:

[0007] The first model parameters are obtained based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system.

[0008] Based on the first model parameters and the mapping relationship, the target parameter values ​​of the PID control system are obtained;

[0009] Based on the target parameter value and the current parameter value of the PID control system, the membership degree of the parameter change of the PID control system relative to at least one fuzzy subset is determined, wherein the at least one fuzzy subset represents at least one rate of change of the parameter value.

[0010] The parameter change rate of the PID control system is determined based on the membership degree of at least one fuzzy subset.

[0011] The current parameter values ​​of the PID control system are updated based on the rate of change of the parameters of the PID control system.

[0012] Optionally, obtaining the first model parameters based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system specifically includes:

[0013] For each candidate model parameter in the at least one candidate model parameter, the error of the candidate model parameter is obtained based on the measured value of the controlled object of the PID control system and the predicted value of the controlled object of the PID control system.

[0014] The first model parameter is determined from the at least one candidate model parameter based on the error of each candidate model parameter.

[0015] Optionally, the error of the candidate model parameters is determined by an objective function, which includes:

[0016]

[0017] Wherein, Δy(k) is the differential signal of the measured value of the controlled object in the PID control system. The predicted value of Δy(k) is given by N, where k is the index number of the differential signal. mdl N is the length of the impulse response sequence stage, d is the number of delay time points, and N is the number of delay time points. DS σ is the maximum value of the index number of the differential signal, p(|K1|) is the log-normal distribution value of the model gain, p(T1) is the log-normal distribution value of the time constant, p(τ1) is the log-normal distribution value of the delay time, σ is the standard deviation of the measurement noise of the measured value of the controlled object of the PID control system, and J is the error of the candidate model parameter.

[0018] Optionally, the method further includes:

[0019] Obtain the log-normal distribution model that the model parameters of the PID control system satisfy;

[0020] Based on the log-normal distribution model, at least one candidate model parameter of the PID control system is obtained.

[0021] Optionally, obtaining at least one candidate model parameter for the PID control system based on the log-normal distribution model specifically includes:

[0022] The model parameters within a pre-defined parameter range are grouped using a grid search method. Each group of model parameters is then substituted into the log-normal distribution model for calculation, and the resulting log-normal distribution values ​​are used as candidate model parameters.

[0023] Optionally, the mapping relationship includes:

[0024]

[0025] Wherein, Kc1 is the proportional parameter value in the target parameter value, Ti1 is the integral parameter value in the target parameter value, Td1 is the differential parameter value in the target parameter value, T1 is the time constant in the first model parameter, and τ1 is the delay time in the first model parameter.

[0026] Optionally, before obtaining the first model parameters based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system, the method further includes:

[0027] The value of the model learning cycle counter is determined to be equal to a pre-set threshold.

[0028] Optionally, updating the current parameter values ​​of the PID control system based on the rate of change of the parameters of the PID control system includes:

[0029] In response to detecting that the difference between the target parameter value and the current parameter value is greater than a step size threshold, the current parameter value is updated to the sum of the current parameter value and the parameter change rate;

[0030] Alternatively, in response to detecting that the difference between the target parameter value and the current parameter value is less than or equal to the step size threshold, the current parameter value is updated to the target parameter value.

[0031] Optionally, after updating the current parameter values ​​of the PID control system based on the rate of change of the parameters of the PID control system, the method further includes:

[0032] The controlled object is controlled based on the updated parameter values ​​of the PID control system.

[0033] Secondly, this application provides a PID control system, comprising:

[0034] The model learning module is used to obtain first model parameters based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system; obtain the target parameter value of the PID control system using the first model parameter and the mapping relationship; determine the membership degree of the parameter change of the PID control system relative to at least one fuzzy subset based on the target parameter value of the PID control system and the current parameter value of the PID control system; and determine the parameter change rate of the PID control system based on the membership degree of at least one fuzzy subset.

[0035] The update module is used to update the current parameter values ​​of the PID control system using the rate of change of the parameters of the PID control system.

[0036] Optionally, the update module is specifically used for:

[0037] Determine whether the difference between the current parameter value and the target parameter value is greater than a preset threshold;

[0038] If it is greater than, then the current parameter value is updated to the sum of the current parameter value and the parameter change rate;

[0039] If the current parameter value is less than or equal to the target parameter value, then the current parameter value is updated to the target parameter value.

[0040] Compared with the prior art, this application has the following beneficial effects:

[0041] In this application, first model parameters are obtained by optimizing at least one candidate model parameter. Based on the mapping relationship between the first model parameters and the target parameter values ​​of the PID parameters, the target parameter values ​​of the PID parameters are calculated. Using fuzzy theory, the rate of change of the PID parameters is calculated. The difference between the current parameter value and the target parameter value, along with the magnitude of the rate of change, is determined, and the PID parameters are then updated. In this way, the parameters of the PID control system can be adjusted in a timely manner according to changes in the dynamic characteristics of the industrial production process. Therefore, the stability of the closed-loop system of the PID control system can be improved. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 A flowchart illustrating a method for parameter self-adjustment of a PID control system provided in an embodiment of this application;

[0044] Figure 2 Another flowchart of a parameter self-adjustment method for a PID control system provided in this application embodiment;

[0045] Figure 3 A line graph showing the input fuzzy quantity of a method for parameter self-adjustment of a PID control system provided in an embodiment of this application;

[0046] Figure 4A line graph showing the output fuzzy quantity of a parameter self-adjustment method for a PID control system provided in this application embodiment;

[0047] Figure 5 This is a schematic diagram of the structure of a PID control system provided in an embodiment of this application. Detailed Implementation

[0048] To enable those skilled in the art to better understand the technical solution of this application, the application scenario of this application will be described first below.

[0049] Stable control of industrial production processes is fundamental to industrial process control. Traditional PID control systems, whose parameters remain unchanged after setting, can meet the control requirements of most industrial processes. Furthermore, traditional PID control is relatively simple and easy to operate, leading to its widespread adoption in industry. Currently, over 80% of control loops in industry utilize this simple PID control system. However, in actual production processes, the dynamic characteristics of the equipment are constantly changing due to various factors such as environmental disturbances, equipment load, and catalyst deactivation.

[0050] Currently, traditional PID control systems cannot automatically adjust parameters according to the dynamic characteristics of industrial production equipment. Alternatively, they require on-site testing or preset operating conditions to set parameters. Furthermore, the degree of change in dynamic characteristics may exceed the anti-interference capability of the PID control system, potentially affecting the timeliness of parameter changes when the deviation is significant. Therefore, this can lead to instability in the closed-loop system, slow response to parameter changes, and a limited range of applications.

[0051] Multiple parameters, engineering tuning, fuzzy PID, and neural network technology are existing PID control methods that can adapt to dynamic changes in on-site production.

[0052] Multiple parameter methods can pre-define operating conditions and design PID control system parameters separately for each condition. However, production often operates under specific conditions, and if unforeseen special conditions arise, the methods lack guidance. Engineering tuning methods require prior open-loop testing or closed-loop testing under proportional control, placing certain requirements on field testing conditions and making them unsuitable for environments with high safety requirements or where signal testing is prohibited. Fuzzy PID input methods control error and its rate of change, but error, as feedback information, exhibits lag; when dynamic characteristics change significantly, the control system parameters struggle to respond quickly. Neural network techniques are used to output PID control system parameters. Considering computational time, offline training of the neural network is employed; online training only performs forward calculations. All effects are affected by the data range; if data is acquired under specific operating conditions, the neural network technique's generalization ability under other conditions is insufficient. However, all four methods mentioned above suffer from closed-loop system instability, slow parameter change response speed, and limited applicability.

[0053] To address the aforementioned technical problems, this application provides a method and system for parameter self-adjustment of a PID control system. In this application's technical solution, first model parameters are obtained by optimizing at least one candidate model parameter. Based on the mapping relationship between the first model parameters and the target parameter values ​​of the PID parameters, the target parameter values ​​of the PID parameters are calculated. Using fuzzy theory, the rate of change of the PID parameters is calculated. The difference between the current parameter value and the target parameter value, along with the magnitude of the rate of change, is determined, and the PID parameters are then updated. In this way, the parameters of the PID control system can be adjusted in a timely manner according to changes in the dynamic characteristics of the industrial production process. Therefore, the stability of the closed-loop system of the PID control system can be improved.

[0054] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0055] Figure 1 This is a flowchart illustrating a method for parameter self-adjustment of a PID control system provided in an embodiment of this application. Figure 1 As shown, the method includes:

[0056] S101: Obtain the first model parameters based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system.

[0057] In PID control, the measured value of the controlled object needs to be adjusted proportionally, integrally, and derivatively to ensure that the measured value reaches the target value. The proportional parameter is denoted as Kc, the integral parameter as Ti, and the derivative parameter as Td.

[0058] Assuming the control process is a first-order process with a pure delay, a candidate model G(s) is constructed, where G(s) is a transfer function. The specific transfer function formula is as follows:

[0059]

[0060] K is the model gain, T is the time constant, and τ is the delay time. K, T, and τ are the first candidate model parameters for PID control.

[0061] Based on the user-defined range of model parameters, the first candidate model parameters can be grouped using a grid search method. Optimization calculations are then performed on each group of first candidate model parameters to obtain second candidate model parameters. The error of the candidate model parameters can be calculated using the measured values ​​of the controlled object in the PID control system and the second candidate model parameters. The group of first candidate model parameters with the smallest error value is then selected as the first model parameters. The first model parameters are identified as K1, T1, and τ1.

[0062] S102: Based on the first model parameters and the mapping relationship, obtain the target parameter values ​​of the PID control system.

[0063] The calculated parameters of the first model have a certain mapping relationship with the parameters of the PID control system. The specific mapping formula is as follows:

[0064]

[0065] Where Kc1 is the proportional parameter value in the target parameter value, Ti1 is the integral parameter value in the target parameter value, Td1 is the differential parameter value in the target parameter value, T1 is the time constant in the first model parameter, and τ1 is the delay time in the first model parameter.

[0066] Using the mapping formula described above, the target parameter values ​​of the PID control system can be obtained.

[0067] S103: Based on the target parameter values ​​and current parameter values ​​of the PID control system, determine the membership degree of the parameter changes of the PID control system relative to at least one fuzzy subset.

[0068] Changes in the parameters of a PID control system will affect the closed-loop system. Therefore, the parameters of a PID control system should not change abruptly but rather gradually. However, if there is a significant difference between the current parameter value and the target parameter value, the rate of parameter change needs to be appropriately increased to prevent closed-loop instability caused by parameter mismatch in the controlled process.

[0069] Furthermore, due to the changing characteristics of the parameters in the PID control system, increasing Kc is detrimental to the stability of the closed-loop system. Therefore, based on the mapping relationship of S102, it can be analyzed that when the first model parameter K1 increases, T1 decreases, or τ1 increases, Kc should reach the target parameter Kc1 more quickly to prevent the closed-loop system from becoming unstable.

[0070] Therefore, the parameter change can be determined using the target parameter value and the current parameter value of the PID control system. According to the calculation rules of fuzzy theory, the parameter change is used as the input fuzzy quantity. A fuzzy subset A of the input fuzzy quantity and a fuzzy subset B of the output fuzzy quantity are designed, and the membership value of at least one fuzzy subset of the parameter change is calculated.

[0071] S104: Determine the rate of change of parameters of the PID control system based on the membership degree of at least one fuzzy subset.

[0072] Based on human experience, fuzzy rule tables for fuzzy subsets A and B are designed. The elements in these tables represent the correspondence rules between the membership degrees of the input and output fuzzy quantities. Based on the correspondences in the fuzzy rule tables, the specific value of the output fuzzy quantity is calculated.

[0073] Based on the mapping relationship between the output fuzzy quantity and the target parameter value and the current parameter value, the parameter change rate of the PID control system can be calculated.

[0074] S105: Update the current parameter values ​​of the PID control system based on the parameter change rate of the PID control system.

[0075] Determine whether the absolute value of the difference between the current parameter value and the target parameter value of the PID control system is greater than the preset minimum step size for parameter change in the PID control system.

[0076] If the value is greater than the target value, the current parameter value is updated to the sum of the current parameter value and the rate of change of the parameter; if the value is less than or equal to the target value, the current parameter value is updated to the target parameter value.

[0077] The updated current parameter values ​​are used to adjust the controlled object of the PID control system and control the output.

[0078] In this application, first model parameters are obtained by optimizing at least one candidate model parameter. Based on the mapping relationship between the first model parameters and the target parameter values ​​of the PID parameters, the target parameter values ​​of the PID parameters are calculated. Using fuzzy theory, the rate of change of the PID parameters is calculated. The difference between the current parameter value and the target parameter value, and the magnitude of the rate of change, are compared with the PID parameter values ​​before updating the PID parameters. Therefore, the stability of the closed-loop system of the PID control system can be improved. Furthermore, model learning can avoid dividing the process industry into different operating conditions, making it widely applicable. Optimizing the model parameters first and then performing fuzzification to obtain the rate of change of parameters can improve the slow response speed to parameter changes.

[0079] Figure 2 Another flowchart illustrating a method for parameter self-tuning of a PID control system provided in this application embodiment. (See attached flowchart.) Figure 2 As shown, the method includes:

[0080] S201: Initialize the model learning cycle counter, the minimum step size of the PID control system parameter change, and the current parameter values ​​of the PID control system.

[0081] The model learning cycle counter is used to record the number of cycles of PID control execution. Users can set the threshold of the model learning cycle counter. When the value of the model learning cycle counter is equal to the preset threshold, steps S203 to S205 need to be executed to perform model learning for PID control.

[0082] Specifically, during the initial power-on operation of the PID control system, it is necessary to initialize the model learning cycle counter, the minimum step sizes of PID control parameter changes (ΔKcmin, ΔTimin, and ΔTdmin), and the current parameter values ​​Kc, Ti, and Td of the PID control system. The model learning cycle counter is set to a value one natural number smaller than a preset threshold. Here, Kc is the proportional parameter, Ti is the integral parameter, and Td is the derivative parameter.

[0083] For example, if the preset threshold for the model learning cycle counter is 8, it means that when the PID control is executed for the eighth cycle, model learning for the PID control needs to be performed to adapt to the constantly changing dynamic characteristics of on-site production. At this time, the initial value of the model learning cycle counter needs to be set to 7, and the minimum step size of the PID control parameter change and the current parameter value of the PID control system need to be set according to human experience.

[0084] S202: Update the model learning cycle counter.

[0085] The model learning cycle counter is incremented by one to update it. It is then determined whether the updated model learning cycle counter is equal to the preset threshold. If it is equal, the process proceeds to S203 to perform the model learning function for PID control. If it is not equal, the process proceeds to S206 to update the current parameter values ​​of the PID control system.

[0086] For example, suppose this is the first run of the PID control system after power-on. The initial value of the model learning cycle counter is 7, and the preset threshold is 8. After updating the model learning cycle counter, the value is 8. If the current value of the model learning cycle counter equals the preset threshold, then the process needs to enter S203 to perform the PID control model learning function. Before performing the model learning function, the value of the model learning cycle counter is reset to zero. Resetting to zero is a common practice in process industries. Therefore, during the loop, there is no need to check whether the model learning cycle counter meets the threshold multiple. If the current value of the model learning cycle counter is 4, and the updated value is 5, then the current value of the model learning cycle counter is not equal to the preset threshold, and the process needs to enter S206 to update the current parameter values ​​of the PID control system.

[0087] S203: Obtain the first model parameters using at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system.

[0088] Suppose a candidate model of a PID control system in the form of a transfer function. Where K is the model gain, T is the time constant, and τ is the delay time, and K, T, and τ are the first candidate model parameters for PID control.

[0089] Users can pre-set the parameter range of the first candidate model based on actual industrial production conditions. The first candidate model parameters are then grouped within the pre-set parameter range using a grid search method.

[0090] Establish a log-normal distribution model for the parameters of the first candidate model. and The parameters of each first candidate model are substituted into the log-normal distribution model for calculation.

[0091] Specifically, users can set the shape parameters of the log-normal distribution model according to industrial production conditions. and Using the formula for calculating the log-normal distribution, the parameters p(|K1|), p(T1), and p(τ1) of the second candidate model are calculated. In other words, the parameters of the second candidate model are the log-normal distribution values ​​of the grouped parameters of the first candidate model. The formula for calculating the log-normal distribution is as follows:

[0092]

[0093]

[0094]

[0095] Based on the objective function J and the second candidate model parameters, the error of the model parameters is calculated. The error values ​​of each group of model parameters are compared, and the group of first candidate model parameters with the smallest error value is taken as the first model parameters K1, T1 and τ1.

[0096] Specifically, the objective function J represents the error between the measured value and the predicted value of the controlled object in a PID control system. The formula for the objective function J is as follows:

[0097]

[0098] Where Δy(k) is the differential signal of the measured value of the controlled object in the PID control system. The predicted value of Δy(k), where k is the index of the differential signal, and N is the predicted value. mdl N is the length of the impulse response sequence stage, d is the number of delay time points, and N is the number of delay time points. DS σ is the maximum value of the index number of the differential signal, p(|K1|) is the log-normal distribution value of the model gain, p(T1) is the log-normal distribution value of the time constant, p(τ1) is the log-normal distribution value of the delay time, σ is the standard deviation of the measurement noise of the measured value of the controlled object in the PID control system, and J is the error of the first candidate model parameter.

[0099] During model learning, it is not necessary to offline segmentation of multiple operating conditions, and it is not limited by unknown operating conditions, thus having a wider range of applications. Furthermore, the log-normal distribution allows the parameters of the first candidate model to fit more accurately with the measured values ​​of the controlled object and the operating controlled object of the PID control system, improving the overfitting problem caused by noise and interference.

[0100] S204: Calculate the target parameter values ​​of the PID control system.

[0101] Using the obtained first model parameters K1, T1, and τ1, and the empirical formula of the response curve method (a common mapping relationship for calculating PID control system parameters), the target parameter values ​​Kc1, Ti1, and Td1 of the PID control system are calculated. Using empirical formulas to calculate the target parameter values ​​of the PID control system can improve the stability of the closed-loop system. The mapping relationship between the empirical formula of the response curve method and the first model parameters is as follows:

[0102]

[0103] Where Kc1 is the proportional parameter value in the target parameter values, Ti1 is the integral parameter value in the target parameter values, Td1 is the derivative parameter value in the target parameter values, T1 is the time constant in the first model parameters, and τ1 is the delay time in the first model parameters. S205: Calculate the parameter change rate of the PID control system.

[0104] Based on the calculation rules of fuzzy theory, the input fuzzy quantities α and β are designed, the universe of discourse is X=(0,∞), and the fuzzy subset A={far-small (RS), small (S), close (N), large (B), far-large (RB)}.

[0105] The input fuzzy quantity is the relationship between the target parameter value and the current parameter value. In this embodiment, the input fuzzy quantity is the ratio between the target parameter value and the current parameter value. It is understood that other calculation methods can also be chosen. The formula for the relationship between the input fuzzy quantities α and β according to fuzzy theory rules is as follows:

[0106]

[0107] The formula for the semantic function of the input fuzzy quantity of the fuzzy theory rule is as follows:

[0108]

[0109] The membership formula for the semantic function expansion of the input fuzzy quantity of the fuzzy theory rule is as follows:

[0110]

[0111]

[0112]

[0113]

[0114]

[0115] The membership degrees of the input fuzzy quantities α and β are as follows: Figure 3 As shown. Substituting the input fuzzy quantities α and β into x in the membership formula yields two fuzzy subsets: fuzzy subset A1 corresponds to input fuzzy quantity α, and fuzzy subset A2 corresponds to input fuzzy quantity β. Calculating the direct product of fuzzy subsets A1 and A2 results in a 5×5 matrix.

[0116] The design outputs a fuzzy quantity γ with a universe of discourse of Z = [0.1, 0.5] and a fuzzy subset B = {slow (SL), medium (M), and fast (FA)}. A fuzzy rule table is set based on human experience, as follows:

[0117] Table 1 Fuzzy Rule Table

[0118]

[0119] The semantic function formula for the output fuzzy quantity of the fuzzy rule is as follows:

[0120]

[0121] The membership formula for the semantic function expansion of the output fuzzy quantity of the fuzzy theory rule is as follows:

[0122]

[0123]

[0124]

[0125] The membership relationship of the output fuzzy quantity γ, such as Figure 4 As shown in Table 1, based on the fuzzy rule table and the membership formula for the output fuzzy quantity, the membership value of the output fuzzy quantity γ can be calculated, which is the result of the fuzzy subset B. Then, based on the fuzzy decision, the specific value of the output fuzzy quantity γ is calculated.

[0126] Based on the mapping relationship between the output fuzzy quantity γ and the target parameter value and the current parameter value, the parameter change rates ΔKc, ΔTi and ΔTd of the PID control system are calculated.

[0127] The mapping relationship between the output fuzzy quantity and the target parameter value and the current parameter value is as follows:

[0128]

[0129] The larger the output fuzzy value γ, the greater the calculated rate of change of the parameter, and the faster the PID control system adjusts its current parameter value towards the target parameter value. Conversely, the smaller the output fuzzy value γ, the smaller the calculated rate of change of the parameter, and the slower the PID control system adjusts its current parameter value towards the target parameter value.

[0130] S206: Update the current parameter values ​​of the PID control system.

[0131] Determine whether the absolute value of the difference between the detected target parameter value and the current parameter value of the PID control system is greater than the preset minimum step size of the parameter change rate. If it is greater, update the current parameter value to the sum of the current parameter value and the parameter change rate obtained in S205; if it is less than or equal to the minimum step size, update the current parameter value to the target parameter value.

[0132] Specifically, it checks whether the current parameter value Kc of the PID control system is equal to the target parameter value Kc1. If they are equal, the current parameter value Kc is maintained; otherwise, it is updated. When |Kc-Kc1|≤Kcmin, the current parameter value Kc is updated to the target parameter value Kc1. When |Kc-Kc1|>Kcmin, the current parameter value Kc is updated to the sum of the current parameter and the rate of change of the parameter, that is, Kc=Kc+△Kc. It then checks whether the current parameter value Ti of the PID control system is equal to the target parameter value Ti1. If they are equal, the current parameter value Ti is maintained; otherwise, it is updated. When |Ti-Ti1|≤Timin, the current parameter value Ti is updated to the target parameter value Ti1. When |Ti-Ti1|>Timin, the current parameter value Ti is updated to the sum of the current parameter and the rate of change of the parameter, that is, Ti=Ti+△Ti. Determine whether the current parameter value Td in the current PID control system is equal to the target parameter value Td1. If they are equal, keep the current parameter value Td. If they are not equal, update Td. When |Td-Td1|≤Tdmin, update the current parameter value Td to the target parameter value Td1. When |Td-Td1|>Tdmin, update the current parameter value Td to the sum of the current parameter and the rate of change of the parameter, that is, Td=Td+△Td.

[0133] By comparing the current parameter values ​​of the PID control system with the target parameter values, and updating the target parameter values ​​in a timely manner based on the dynamic characteristics of the model, and calculating the parameter change rate of the PID control system according to the rules of fuzzy theory, the timeliness of parameter changes in the PID control system and the process fluctuations caused by excessively rapid parameter changes in the PID control system can be satisfied simultaneously.

[0134] S207: Issue the updated PID current parameter values ​​and implement PID control.

[0135] The updated current parameter values ​​are used to adjust the PID control object and calculate the control output. If the adjusted current parameter values ​​still differ from the target parameter values, S206 is repeated in one model learning cycle to update the current parameter values.

[0136] In this application, the need for model learning is determined by cyclically recording values ​​using a model learning cycle counter. When the counter value equals a pre-set threshold, model learning is initiated, optimizing the parameters of the first candidate model to obtain a second candidate model. The model function with the smallest objective function error is selected as the first model parameter. Based on the mapping relationship between the first model parameter and the target parameter value of the PID parameter, the target parameter value of the PID parameter is calculated. Using fuzzy theory, the parameter change rate of the PID parameter is calculated. The difference between the current parameter value and the target parameter value of the PID parameter, along with the magnitude of the parameter change rate, is compared to update the PID parameter. This avoids dividing the process industry into different operating conditions, optimizing the model parameters first, and then performing fuzzy processing to obtain the parameter change rate, thus improving the slow response speed to parameter changes. Therefore, it can improve the stability of the closed-loop system, provide timely response to parameter changes, and has a wide range of applications.

[0137] Figure 5 This is a schematic diagram of a PID control system provided in an embodiment of this application. Figure 5 As shown, the system includes:

[0138] The model learning module 510 is used to obtain first model parameters based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system; obtain target parameter values ​​of the PID control system using the first model parameters and the mapping relationship; determine the membership degree of the parameter change of the PID control system relative to at least one fuzzy subset based on the target parameter values ​​and the current parameter values ​​of the PID control system; and determine the parameter change rate of the PID control system based on the membership degree of at least one fuzzy subset.

[0139] The update module 520 is used to update the current parameter values ​​of the PID control system using the rate of change of the parameters of the PID control system.

[0140] Specifically, the update module is used to determine whether the difference between the current parameter value and the target parameter value is greater than a preset threshold; if it is greater, the current parameter value is updated to the sum of the current parameter value and the parameter change rate; if it is less than or equal to the target parameter value, the current parameter value is updated to the target parameter value.

[0141] In this application, the parameters of the PID control system are adjusted through a model learning module, and then the current parameter values ​​are updated using an update module. This avoids dividing the process industry into operating conditions, optimizes the model parameters first, and then performs fuzzification processing to obtain the parameter change rate, thus improving the slow response speed to parameter changes. Therefore, it can improve the stability of the closed-loop system, provide timely response to parameter changes, and has a wide range of applications.

[0142] In the embodiments of this application, the terms "first" and "second" (if they exist) are used only as name identifiers and do not represent the order of first and second.

[0143] It should be noted that the various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for the device and system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiments. The device and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components indicated as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the solution in this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0144] The above description is merely one specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for parameter self-adjustment of a PID control system, characterized in that, Applied to a PID control system, the method includes: Initialize the model learning cycle counter, the minimum step size for PID control system parameter changes, and the current parameter values ​​of the PID control system; update the model learning cycle counter, and determine whether the updated model learning cycle counter equals the preset threshold. If equal, the first model parameter is obtained based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system; Based on the first model parameters and the mapping relationship, the target parameter values ​​of the PID control system are obtained; Based on the target parameter value and the current parameter value of the PID control system, the membership degree of the parameter change of the PID control system relative to at least one fuzzy subset is determined, wherein the at least one fuzzy subset represents at least one rate of change of the parameter value. The parameter change rate of the PID control system is determined based on the membership degree of at least one fuzzy subset. The current parameter values ​​of the PID control system are updated based on the rate of change of the parameters of the PID control system. The first model parameters are obtained based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system, including: Candidate models for PID control systems in the form of transfer functions are selected. Where K is the model gain and T is the time constant. For the delay time, and K, T and The first candidate model parameter is selected as the PID control parameter; the parameter range of the first candidate model parameter is set, and the first candidate model parameter is grouped within the pre-set parameter range using a grid search method; Establish a log-normal distribution model for the parameters of the first candidate model. , and ; The shape parameters of the log-normal distribution model are set according to industrial production conditions. , and The parameters of the second candidate model were calculated using the log-normal distribution formula. , and The second candidate model parameters are the log-normal distribution values ​​of the grouped first candidate model parameters; the second candidate model parameters , and The probability density values ​​of the model gain K, time constant T, and delay time τ in the first candidate model parameters are respectively represented by the log-normal distribution model. The probability density values ​​are used as penalty terms in the objective function J to constrain the rationality of the values ​​of the first candidate model parameters. The formula for calculating the log-normal distribution is as follows: ； ； ； Based on the objective function J and the second candidate model parameters, the error of the model parameters is calculated. The error values ​​of each group of model parameters are compared, and the group of first candidate model parameters with the smallest error value is selected as the first model parameters. , and ; The step of updating the current parameter values ​​of the PID control system based on the rate of change of the parameters of the PID control system includes: In response to detecting that the difference between the target parameter value and the current parameter value is greater than a step size threshold, the current parameter value is updated to the sum of the current parameter value and the parameter change rate; In response to detecting that the difference between the target parameter value and the current parameter value is less than or equal to the step size threshold, the current parameter value is directly updated to the target parameter value; The at least one fuzzy subset includes five fuzzy subsets: far small, small, close, large, and far large, whose membership function adopts a piecewise linear function; the parameter change is the ratio of the target parameter value to the current parameter value, including... ; The first candidate model parameters are grouped within a pre-set parameter range using a grid search method. Specifically, this includes sampling at equal intervals within the parameter range with a preset step size to generate a finite number of first candidate model parameter groups. The step size is set by the user based on the parameter identification accuracy.

2. The method according to claim 1, characterized in that, The first model parameters are obtained based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system, specifically including: For each candidate model parameter in the at least one candidate model parameter, the error of the candidate model parameter is obtained based on the measured value of the controlled object of the PID control system and the predicted value of the controlled object of the PID control system. The first model parameter is determined from the at least one candidate model parameter based on the error of each candidate model parameter.

3. The method according to claim 2, characterized in that, The error of the candidate model parameters is determined by an objective function, which includes: ； in, It is the differential signal of the measured value of the controlled object in the PID control system. For the The predicted value, The index number of the differential signal. The length of the impulse response sequence phase. The number of time points to delay. , The log-normal distribution value of the model gain. The time constant is a log-normal distribution value. The delay time is a log-normal distribution value. The standard deviation of the measurement noise of the measured value of the controlled object in the PID control system is given. The error is the parameter of the candidate model.

4. The method according to claim 1, characterized in that, The method further includes: Obtain the log-normal distribution model that the model parameters of the PID control system satisfy; Based on the log-normal distribution model, at least one candidate model parameter of the PID control system is obtained.

5. The method according to claim 4, characterized in that, The step of obtaining at least one candidate model parameter for the PID control system based on the log-normal distribution model specifically includes: The model parameters within a pre-defined parameter range are grouped using a grid search method. Each group of model parameters is then substituted into the log-normal distribution model for calculation, and the resulting log-normal distribution values ​​are used as candidate model parameters.

6. The method according to claim 1, characterized in that, The mapping relationship includes: ； in, The proportional parameter value in the target parameter value. The integral parameter value in the target parameter value. The differential parameter value in the target parameter value. The time constant in the parameters of the first model. The delay time is the parameter in the first model.

7. The method according to claim 1, characterized in that, Before obtaining the first model parameters based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system, the method further includes: The value of the model learning cycle counter is determined to be equal to a pre-set threshold.

8. The method according to claim 1, characterized in that, After updating the current parameter values ​​of the PID control system based on the rate of change of the parameters of the PID control system, the method further includes: The controlled object is controlled based on the updated parameter values ​​of the PID control system.

9. A PID control system, characterized in that, A method for performing parameter self-tuning of a PID control system according to any one of claims 1-8, comprising: The model learning module is used to initialize the model learning cycle counter, the minimum step size of the PID control system parameter change, and the current parameter value of the PID control system; update the model learning cycle counter, determine whether the updated model learning cycle counter is equal to a preset threshold, and if so, obtain the first model parameter based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system; obtain the target parameter value of the PID control system using the first model parameter and the mapping relationship; determine the membership degree of the parameter change of the PID control system relative to at least one fuzzy subset based on the target parameter value and the current parameter value of the PID control system; and determine the parameter change rate of the PID control system based on the membership degree of at least one fuzzy subset. The step of obtaining the first model parameter based on at least one candidate model parameter of the PID control system and the measured value of the controlled object of the PID control system includes: Candidate models for PID control systems in the form of transfer functions are selected. Where K is the model gain and T is the time constant. For the delay time, and K, T and The first candidate model parameter is selected as the PID control parameter; the parameter range of the first candidate model parameter is set, and the first candidate model parameter is grouped within the pre-set parameter range using a grid search method; Establish a log-normal distribution model for the parameters of the first candidate model. , and ; The shape parameters of the log-normal distribution model are set according to industrial production conditions. , and The parameters of the second candidate model were calculated using the log-normal distribution formula. , and The second candidate model parameters are the log-normal distribution values ​​of the grouped first candidate model parameters; the formula for calculating the log-normal distribution is as follows: ； ； ； Based on the objective function J and the second candidate model parameters, the error of the model parameters is calculated. The error values ​​of each group of model parameters are compared, and the group of first candidate model parameters with the smallest error value is selected as the first model parameters. , and ; The update module is used to update the current parameter values ​​of the PID control system using the rate of change of the parameters of the PID control system.

10. The system according to claim 9, characterized in that, The update module is specifically used for: Determine whether the difference between the current parameter value and the target parameter value is greater than a preset threshold; If it is greater than, then the current parameter value is updated to the sum of the current parameter value and the parameter change rate; If the current parameter value is less than or equal to the target parameter value, then the current parameter value is updated to the target parameter value.

Images