A method and system for intelligent tuning of PID loop parameters based on key indicators

Abstract

This invention discloses a method and system for intelligent tuning of PID loop parameters based on key indicators, relating to the field of automated PID control. The method includes: collecting operating parameters; setting key indicators reflecting control performance; calculating the actual values ​​of the key indicators and setting thresholds; determining whether the PID loop meets the control performance degradation requirements; if so, filtering modeling data segments; performing model identification on the PID loop; tuning the PID loop parameters and inputting the suggested values ​​of the PID parameters into the distributed control system (DCS); continuing to collect operating parameters and calculate the actual values ​​of the key indicators; if not, directly continuing to collect operating parameters and calculate the actual values ​​of the key indicators. This invention judges the control performance of the PID loop based on the relationship between the actual values ​​and thresholds, thereby enabling timely intelligent parameter tuning when the control performance of the PID loop deteriorates or the control effect is poor, ensuring that the control loop always operates in an optimal state.

Description

Technical Field

[0001] This invention relates to the field of automated PID control, and more specifically, to a method and system for intelligent tuning of PID loop parameters based on key indicators. Background Technology

[0002] PID (Proportional Integral Derivative) control is the most widely used control strategy in industrial controllers due to its simple structure, convenient parameter adjustment, wide applicability, and strong robustness. The performance of PID loop control plays a crucial role in improving production efficiency, ensuring safe and stable operation, and reducing costs, directly impacting the economic benefits of enterprises.

[0003] The biggest challenge in PID control loops lies in PID parameter tuning. Currently, PID parameters are mostly tuned based on engineers' experience, which has many drawbacks. For example, firstly, manual tuning relies heavily on engineers' experience and is inefficient; even experienced personnel need to spend a significant amount of time on engineering testing and parameter optimization. Secondly, engineers can only qualitatively judge the quality of PID parameters through curves displayed on the DCS (Distributed Control System) screen, and cannot accurately measure them quantitatively. Thirdly, a set of PID parameters has a limited range of applicability. When production conditions such as load change, causing a shift in operating state, the original PID parameters may no longer be applicable, leading to a decline in the control performance of the PID controller. Engineers often struggle to accurately detect this shift in operating state and adjust the PID parameters in a timely manner, resulting in the PID controller not always operating at its optimal state.

[0004] Therefore, designing an intelligent method for automatically tuning PID parameters to achieve optimal control performance of the PID controller is a technical problem that urgently needs to be solved in the industry. Summary of the Invention

[0005] This invention provides a method and system for intelligent tuning of PID loop parameters based on key indicators, in order to overcome at least one technical problem existing in the prior art.

[0006] On one hand, embodiments of the present invention provide a method for intelligent tuning of PID loop parameters based on key indicators, including:

[0007] The operating parameters of the PID loop are acquired from the distributed control system (DCS).

[0008] Define key indicators that reflect the control performance of the PID loop;

[0009] Calculate the actual values ​​of the key indicators based on the operating parameters of the PID loop;

[0010] Based on the characteristics of the PID loop, set the threshold values ​​for the key indicators;

[0011] Based on the relationship between the actual value and the threshold, determine whether the PID loop meets the control performance degradation requirement. If yes: send a control performance degradation notification and activate the intelligent PID parameter tuning function; select data segments from the operating parameters that reflect the characteristics of the PID loop and are suitable for modeling; perform model identification on the PID loop based on the data segments to obtain the loop model; perform parameter tuning on the PID loop based on the loop model and the type of the distributed control system (DCS) to obtain suggested values ​​for the PID parameters; input the suggested values ​​for the PID parameters into the DCS; continue to collect the operating parameters of the PID loop and calculate the actual values ​​of the key indicators; if no: directly continue to collect the operating parameters of the PID loop and calculate the actual values ​​of the key indicators.

[0012] Optionally, the operating parameters of the PID loop include at least n measured values ​​PV within a set time period. i and n setpoints SV i The key indicators include the coefficient of variation (Cov) and the accuracy coefficient (Accu); the actual values ​​of the key indicators include the calculated values ​​of the coefficient of variation (Cov) and the calculated values ​​of the accuracy coefficient (Accu).

[0013] Based on the operating parameters of the PID loop, the actual values ​​of the key indicators are calculated, specifically as follows:

[0014] According to the formula Calculate the coefficient of variation Cov, where Ave represents the n measured values ​​PV. i The mean, where Std represents the standard deviation; the mean Ave is... The standard deviation Std is

[0015] According to the formula Calculate the accuracy coefficient Accu.

[0016] Optionally, the threshold includes a threshold for the coefficient of variation (Cov) and a threshold for the accuracy coefficient (Accu);

[0017] The determination of whether the PID loop meets the control performance degradation requirement is based on the relationship between the actual value and the threshold. Specifically:

[0018] The calculated value of the coefficient of variation (Cov) and the threshold, and the calculated value of the accuracy coefficient (Accu) and the threshold are compared respectively. If the calculated value of the coefficient of variation (Cov) is greater than the threshold of the coefficient of variation (Cov) and the calculated value of the accuracy coefficient (Accu) is less than the threshold of the accuracy coefficient (Accu), then the control performance degradation requirement is met; otherwise, the control performance degradation requirement is not met.

[0019] Optionally, the threshold for the coefficient of variation Cov is 2, and the threshold for the accuracy coefficient Accu is 0.8.

[0020] Optionally, the operating parameters of the PID loop include the set value, measured value, and controller output value within a set time period;

[0021] The data segments that reflect the characteristics of the PID loop and are suitable for modeling are selected from the operating parameters, specifically:

[0022] Trend extraction is performed on the measured values ​​and the controller output values ​​to obtain trend extraction results;

[0023] Based on the trend extraction results, a reasonableness analysis is performed on each data point to obtain the reasonableness of each data point;

[0024] By using a sliding window method, the set value, the measured value, and the controller output value are divided into multiple sliding data segments;

[0025] Based on the rationality of each data point and the sliding data segment, the average rationality of all data points in the sliding data segment is obtained and denoted as the rationality of the sliding data segment.

[0026] When the automatic state within the sliding data segment is automatic and the set value remains unchanged, the rationality of the sliding data segment is set to zero.

[0027] The sliding data segment whose reasonableness is a first value is taken as the data segment suitable for modeling.

[0028] Optionally, based on the fitting results, a reasonableness analysis is performed on each data point to obtain the reasonableness of each data point, specifically as follows:

[0029] When the system gain is positive, within a predetermined range, if the controller output value or the measured value remains unchanged, the reasonableness is zero; if the trend extraction result of the measured value and the trend extraction result of the controller output value have the same trend, the reasonableness is positive; otherwise, the reasonableness is negative. When the system gain is negative, within a predetermined range, if the controller output value or the measured value remains unchanged, the reasonableness is zero; if the trend extraction result of the measured value and the trend extraction result of the controller output value have the same trend, the reasonableness is negative; otherwise, the reasonableness is positive.

[0030] Optionally, model identification is performed on the PID loop to obtain a loop model, specifically as follows:

[0031] The PID loop was modeled using the least squares principle, resulting in the following loop model: Where K represents the gain of the PID loop, T is the time constant, and τ represents the pure time delay.

[0032] Optionally, the parameters of the PID loop are tuned, specifically as follows:

[0033] The ZN method is used to tune the parameters of the PID loop.

[0034] On the other hand, the present invention also provides a PID loop parameter intelligent tuning system based on key indicators, comprising:

[0035] The acquisition module is used to acquire the operating parameters of the PID loop from the distributed control system (DCS).

[0036] The setting module is used to set key indicators that reflect the control performance of the PID loop.

[0037] The first calculation module is used to calculate the actual value of the key indicator based on the operating parameters of the PID loop.

[0038] The setting module is used to set the threshold of the key indicators according to the characteristics of the PID loop;

[0039] The judgment module is used to determine whether the PID loop meets the control performance degradation requirement based on the relationship between the actual value and the threshold. If yes: a control performance degradation notification is sent, and the intelligent PID parameter tuning function is activated; data segments that reflect the characteristics of the PID loop and are suitable for modeling are selected from the operating parameters; the PID loop is modeled based on the data segments to obtain a loop model; the PID loop is parameter-tuned based on the loop model and the type of the distributed control system (DCS) to obtain suggested values ​​for the PID parameters; the suggested values ​​for the PID parameters are input into the DCS; the operating parameters of the PID loop are collected again, and the actual values ​​of the key indicators are calculated; if no: the operating parameters of the PID loop are collected again, and the actual values ​​of the key indicators are calculated.

[0040] Optionally, the operating parameters of the PID loop include the setpoint, measured value, and controller output value within a set time period; the system further includes:

[0041] The extraction module is used to extract trends from the measured values ​​and the controller output values ​​to obtain trend extraction results;

[0042] The analysis module performs a reasonableness analysis on each data point based on the trend extraction results to obtain the reasonableness of each data point;

[0043] The segmentation module is used to divide the set value, the measured value, and the controller output value into multiple sliding data segments using a sliding window method.

[0044] The second calculation module is used to obtain the average value of the rationality of all data points in the sliding data segment based on the rationality of each data point and the sliding data segment, and denoted as the rationality of the sliding data segment.

[0045] The calibration module is used to set the rationality of the sliding data segment to zero when the automatic state within the sliding data segment is automatic and the set value remains unchanged.

[0046] The selection module is used to select the sliding data segments whose reasonableness is a first value as suitable data segments for modeling.

[0047] The innovative aspects of this invention include:

[0048] 1. In this embodiment, key indicators that can reflect control performance are set. By calculating the actual value of the key indicators and judging the control performance of the PID loop based on the relationship between the actual value and the threshold, the parameters can be intelligently tuned in a timely manner when the control performance of the PID loop decreases or the control effect is poor. This not only keeps the control loop working in the best state, but also effectively reduces the operation intensity and difficulty of the operator. This is one of the innovations of this embodiment.

[0049] 2. In this embodiment, by repeatedly collecting operating parameters and calculating the actual values ​​of key indicators, changes in loop characteristics caused by changes in production conditions such as load can be detected at any time. This enables timely intelligent parameter tuning during work state transitions, which helps the control loop to always operate in the optimal state. It can also effectively reduce the operational intensity and difficulty for operators, which is one of the innovative points of this embodiment. Attached Figure Description

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 A flowchart of a tuning method provided in an embodiment of the present invention;

[0052] Figure 2 A flowchart of a data filtering segment provided in an embodiment of the present invention;

[0053] Figure 3 This is a schematic diagram of a tuning system provided in an embodiment of the present invention;

[0054] Figure 4 This is another structural schematic diagram of the tuning system provided in an embodiment of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] It should be noted that the terms "comprising" and "having," and any variations thereof, in the embodiments and drawings of this invention are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0057] This invention discloses a method and system for intelligent tuning of PID loop parameters based on key indicators. These will be described in detail below.

[0058] Figure 1 A flowchart of a tuning method provided in an embodiment of the present invention is provided below. Figure 1 The intelligent tuning method for PID loop parameters based on key indicators provided in this implementation includes:

[0059] Step 1: Acquire the operating parameters of the PID loop from the Distributed Control System (DCS);

[0060] Step 2: Set key indicators that reflect the control performance of the PID loop;

[0061] Step 3: Calculate the actual values ​​of key indicators based on the operating parameters of the PID loop;

[0062] Step 4: Set thresholds for key indicators based on the characteristics of the PID loop;

[0063] Step 5: Determine whether the PID loop meets the control performance degradation requirement based on the relationship between the actual value and the threshold. If yes: Send a control performance degradation notification and activate the intelligent PID parameter tuning function; filter data segments from the operating parameters that reflect the characteristics of the PID loop and are suitable for modeling; perform model identification on the PID loop based on the data segments to obtain the loop model; perform parameter tuning on the PID loop based on the loop model and the type of distributed control system (DCS) to obtain suggested values ​​for the PID parameters; input the suggested values ​​for the PID parameters into the DCS; continue to collect the operating parameters of the PID loop and calculate the actual values ​​of key indicators. If no: directly continue to collect the operating parameters of the PID loop and calculate the actual values ​​of key indicators.

[0064] For details, please refer to Figure 1 The intelligent tuning method for PID loop parameters based on key indicators provided in this implementation first collects the operating parameters of the PID loop from the distributed control system (DCS) in step 1. The operating parameters collected here can be selected as needed, such as including but not limited to the setpoint SV, measured value PV, and controller output value OP. In addition, it also includes the proportional coefficient P, integral time I, and derivative time D of the PID controller.

[0065] After acquiring the operating parameters of the PID loop, key indicators reflecting the control performance of the PID loop are set in step 2, and the actual values ​​of the key indicators are calculated in step 3 based on the operating parameters of the PID loop. Here, key indicators can be, for example, the coefficient of variation and the accuracy coefficient. The coefficient of variation can evaluate the characteristics of the system recovering to steady state after deviating from steady state due to external disturbances, while the accuracy coefficient can evaluate the control accuracy between the measured value and the set value of the controlled variable during the process of the system reaching steady state.

[0066] In step 4, based on the characteristics of the PID loop, thresholds for key indicators are set. These thresholds are then used to determine whether the calculated actual values ​​reflect that the loop is maintaining optimal operating conditions. The characteristics of the PID loop here could include, for example, the loop's importance or the characteristics of interference.

[0067] It should be noted that steps 3 and 4 are only for illustrative purposes to show that the tuning method includes two steps: calculating the actual value of the key indicator and setting the threshold of the key indicator. They are not intended to limit the actual execution order. In other embodiments, the threshold can be set first and then the actual value can be calculated. This application does not limit this.

[0068] After obtaining the actual values ​​and thresholds of the key indicators, step 5 compares the magnitudes of the actual values ​​and thresholds. Based on this relationship, it is determined whether the PID loop meets the requirements for decreased control performance. If the PID loop meets these requirements, it indicates that the control loop is not operating in its optimal state, and a notification of decreased control performance is sent, activating the intelligent PID parameter tuning function. During intelligent parameter tuning, data segments that reflect the characteristics of the PID loop and are suitable for modeling are first selected from the collected operating parameters. Based on the selected data segments, the PID loop is modeled; for example, the least squares principle can be used for model identification to obtain the loop model.

[0069] After obtaining the loop model, the PID loop parameters are tuned according to the loop model and the type of Distributed Control System (DCS), yielding suggested values ​​for the PID parameters. The classic ZN empirical tuning method can be used for parameter tuning, and then corresponding PID parameters are given based on different DCS models. For example, when the DCS model is a Supcon ECS700, if the PID parameters are selected in PI form, then... T i =3.3τ,T d =0; if the PID parameter is selected in PID form, then T i =2.0τ,T d =0.5τ. Where P represents the proportionality constant, and T iT represents the integration time. d Let T represent the derivative time, K represent the gain of the PID loop, T be the time constant, and τ represent the pure time delay.

[0070] After obtaining the suggested values ​​of the PID parameters, input the suggested values ​​of the PID parameters into the distributed control system (DCS), and then return to step 1 to collect the operating parameters of the PID loop again and calculate the actual values ​​of key indicators. By repeating the above steps, the control performance of the PID loop is judged periodically. When the control performance of the PID loop declines, its parameters can be intelligently tuned in a timely manner, so that the control loop always works in the best state.

[0071] When the PID loop fails to meet the requirements for degraded control performance, it indicates that the control loop is operating at its optimal state, and parameter tuning is unnecessary. Therefore, return directly to step 1 to continue collecting the PID loop's operating parameters and calculating the actual values ​​of key indicators. By repeatedly executing the above steps, the control performance of the PID loop is periodically assessed. When the control performance of the PID loop degrades or the control effect is poor, intelligent parameter tuning can be performed in a timely manner, ensuring that the control loop always operates at its optimal state.

[0072] The present invention provides a method for intelligent tuning of PID loop parameters based on key indicators. This method sets key indicators that reflect control performance, calculates the actual values ​​of these key indicators, and judges the control performance of the PID loop based on the relationship between the actual values ​​and thresholds. This allows for timely intelligent parameter tuning when the control performance of the PID loop deteriorates or the control effect becomes poor. This not only ensures the control loop operates at its optimal state but also effectively reduces the operational intensity and difficulty for operators. Furthermore, by repeatedly collecting operating parameters and calculating the actual values ​​of key indicators, changes in loop characteristics caused by variations in production conditions such as load can be detected at any time. This enables timely intelligent parameter tuning during shifts in operating states, further ensuring the control loop operates at its optimal state.

[0073] Optionally, the operating parameters of the PID loop shall include at least n measured values ​​PV within a set time period. i and n setpoints SV i Key metrics include the coefficient of variation (Cov) and the accuracy coefficient (Accu); the actual values ​​of these key metrics include the calculated values ​​of the Covariance (Cov) and the Accuracy Coefficient (Accu). Please refer to [reference needed]. Figure 1 In step 3, the actual values ​​of key indicators are calculated based on the operating parameters of the PID loop, specifically according to the formula: Calculate the coefficient of variation Cov, where Ave represents the n measurements PV. i The mean, where Std represents the standard deviation; the mean Ave is... Standard deviation Std is According to the formula Calculate the accuracy coefficient Accu.

[0074] Specifically, in this embodiment, the key indicators include the coefficient of variation (Cov) and the accuracy coefficient (Accu). The value of the coefficient of variation (Cov) is related to the measured value, and the value of the accuracy coefficient (Accu) is related to both the measured value and the set value. Therefore, in this embodiment, the operating parameters of the PID loop include n measured values ​​(PV) within a set time period. i and n setpoints SV i .

[0075] Please refer to Figure 1 The actual values ​​of key indicators include the calculated values ​​of the coefficient of variation (Cov) and the coefficient of accuracy (Accu). Therefore, calculating the actual values ​​of key indicators includes calculating the calculated values ​​of the coefficient of variation (Cov) and the coefficient of accuracy (Accu).

[0076] The coefficient of variation can assess the characteristics of a system that, after being deviating from its steady state due to external disturbances, recovers to a steady state. In this embodiment, measured values ​​within a set time period are selected and denoted as PV1, PV2, PV3, ..., PV i ..., PVn, based on the measured value PV i The n measured values ​​PV can be calculated. i The mean is Standard deviation is Then, based on the mean and standard deviation, the coefficient of variation can be calculated as follows:

[0077] The accuracy coefficient can evaluate the control accuracy characteristics between the measured value and the setpoint of the controlled variable during the steady-state process of the system. In this embodiment, the setpoints within a set time are collected from the DCS and denoted as SV1, SV2, SV3, ... SV i , ..., SVn, based on the set value and the measured value, the accuracy coefficient can be calculated as follows:

[0078] In this embodiment, the coefficient of variation and the accuracy coefficient are set as key indicators. The coefficient of variation can evaluate the characteristics of the system to recover to steady state after deviating from steady state due to external disturbances. The accuracy coefficient can evaluate the control accuracy between the measured value and the set value of the controlled variable during the process of the system reaching steady state. By calculating the actual values ​​of the coefficient of variation and the accuracy coefficient and comparing the relationship between the actual values ​​and the threshold, the control performance of the PID loop can be determined. Thus, when the control performance of the PID loop decreases or the control effect is poor, the parameters can be intelligently tuned in time, so that the control loop always works in the best state.

[0079] Optionally, the thresholds include a threshold for the coefficient of variation (Cov) and a threshold for the accuracy coefficient (Accu); please refer to [reference needed]. Figure 1 In step 5, the control performance degradation requirement is determined based on the relationship between the actual value and the threshold. Specifically, the calculated value of the coefficient of variation Cov and the threshold, and the calculated value of the accuracy coefficient Accu and the threshold are compared respectively. If the calculated value of the coefficient of variation Cov is greater than the threshold of the coefficient of variation Cov and the calculated value of the accuracy coefficient Accu is less than the threshold of the accuracy coefficient Accu, then the control performance degradation requirement is met; otherwise, the control performance degradation requirement is not met.

[0080] For details, please refer to Figure 1 This invention determines whether a PID loop meets the requirements for degraded control performance by judging the relationship between the actual values ​​of key indicators and threshold values. Key indicators include the coefficient of variation and the accuracy coefficient. Therefore, in the judgment, the calculated value of the coefficient of variation and the threshold value, as well as the calculated value of the accuracy coefficient and the threshold value, are compared. When both requirements are met simultaneously, it indicates that the control performance of the PID loop has degraded, and intelligent parameter tuning is required.

[0081] The coefficient of variation (Cov) assesses the system's ability to recover to a steady state after being deviated from it by external disturbances. The accuracy coefficient assesses the control accuracy between the measured and setpoint values ​​of the controlled variable during the process of reaching steady state. In a PID control loop, a lower Cov and a higher accuracy coefficient indicate better control performance. Therefore, in this embodiment, when the calculated Cov value is greater than a threshold and the calculated accuracy coefficient (Accu) is less than a threshold, the control performance degradation requirement is met, and intelligent parameter tuning of the PID loop is performed. If either the relationship between the calculated Cov value and the threshold, or the relationship between the calculated accuracy coefficient and the threshold, does not meet the above requirements, the control performance degradation requirement is not met, and intelligent parameter tuning is not performed.

[0082] This invention judges the control performance of the PID loop based on the relationship between the actual value of key indicators and the threshold value. This allows for timely intelligent parameter tuning when the control performance of the PID loop declines or the control effect is poor, ensuring that the control loop always operates in the optimal state.

[0083] Optionally, the threshold for the coefficient of variation (Cov) is 2, and the threshold for the accuracy coefficient (Accu) is 0.8. Specifically, in this embodiment, the threshold for the coefficient of variation (Cov) is set to 2, and the threshold for the accuracy coefficient (Accu) is set to 0.8. However, in other embodiments, the thresholds for the coefficient of variation and the accuracy coefficient can be set to other values. For example, for critical PID loops, or loops with stringent performance requirements where even a slight decrease in performance necessitates parameter retuning, the threshold for the coefficient of variation can be set smaller, such as 1.5, and the threshold for the accuracy coefficient (Accu) can be set larger, such as 0.9. This makes it easier to meet the requirements for control performance degradation, thereby making it easier to trigger a control performance degradation notification and initiate intelligent tuning of the PID parameters.

[0084] It should be noted that the thresholds for the coefficient of variation and accuracy coefficient are set to 2, 0.8 or 1.5, 0.9, which are merely illustrative and not intended to limit this application. In actual use, they can be set as needed.

[0085] Optionally, the operating parameters of the PID loop include the set value, measured value, and controller output value within the set time period; Figure 2 A flowchart of a data filtering segment provided in an embodiment of the present invention is shown below. Figure 1 and Figure 2 In step 5, data segments that reflect the characteristics of the PID loop and are suitable for modeling are selected from the operating parameters. Specifically: Step 51, trend extraction is performed on the measured values ​​and controller output values ​​to obtain trend extraction results; Step 52, based on the trend extraction results, a rationality analysis is performed on each data point to obtain the rationality of each data point; Step 53, the setpoint, measured values, and controller output values ​​are divided using a sliding window method to obtain multiple sliding data segments; Step 54, based on the rationality of each data point and the sliding data segments, the average rationality of all data points in the sliding data segment is obtained and recorded as the rationality of the sliding data segment; Step 55, when the automatic state in the sliding data segment is automatic and the setpoint does not change, the rationality of the sliding data segment is set to zero; Step 56, the sliding data segment with the first rationality value is selected as the data segment suitable for modeling.

[0086] For details, please refer to Figure 1 and Figure 2 When selecting data segments from operating parameters that reflect the characteristics of the PID loop and are suitable for modeling, the first step is to extract trends from the measured values ​​and controller output values ​​in step 51 to obtain the trend extraction results. Here, binomial fitting and dynamic programming can be used for trend extraction. Appropriate numbers of segments and segmentation points are selected, and the corresponding binomial fitting results are used as the final trend extraction results.

[0087] After trend fitting, each data point corresponds to a reasonableness score. Therefore, after obtaining the trend extraction results, in step 52, a reasonableness analysis is performed on each data point based on the trend extraction results to obtain the reasonableness score of each data point. Then, in step 53, a sliding window method is used to divide the setpoint, measured value, and controller output value to obtain multiple sliding data segments. When using a sliding window for division, the window width parameter can be determined according to the loop type. For example, for flow and pressure loops, the window width parameter is set to 133, while for temperature and level loops, the window width parameter is set to 800. Then, the segmentation points for window division are determined using the width parameter, thereby determining each sliding data segment.

[0088] In step 54, based on the reasonableness of each data point and the sliding data segment, the average reasonableness of all data points in each sliding data segment can be calculated and denoted as the reasonableness of the sliding data segment. After obtaining the reasonableness of each sliding data segment, in step 55, based on the automatic status and set value within the sliding data segment, it is determined whether reasonableness correction is required. When the automatic status within the sliding data segment is automatic and the set value remains unchanged, it indicates that correction is required, and the reasonableness of the corresponding sliding data segment is set to zero. Otherwise, no correction is required, and the process proceeds directly to the next step.

[0089] After obtaining the reasonableness of each sliding data segment, in step 56, the sliding data segment with the highest reasonableness value is selected as the suitable data segment for modeling. Here, the value with the highest reasonableness of the sliding data segment is taken as the first value.

[0090] Alternatively, please refer to Figure 1 In step 5, based on the fitting results, a reasonableness analysis is performed on each data point to obtain the reasonableness of each data point. Specifically: when the system gain is positive, within a predetermined range, if the controller output value or the measured value remains unchanged, the reasonableness is zero; if the trend extraction result of the measured value and the trend extraction result of the controller output value change in the same direction, the reasonableness is positive; otherwise, the reasonableness is negative. When the system gain is negative, within a predetermined range, if the controller output value or the measured value remains unchanged, the reasonableness is zero; if the trend extraction result of the measured value and the trend extraction result of the controller output value change in the same direction, the reasonableness is negative; otherwise, the reasonableness is positive.

[0091] For details, please refer to Figure 1The reasonableness of each data point depends not only on the measured value and the controller output value, but also on the system gain. Therefore, when performing reasonableness analysis on each data point, the system gain is first determined. When the system gain is positive, within a predetermined range, if the controller output value or the measured value remains unchanged, the reasonableness is zero; if the trend extraction results of the measured value and the trend extraction results of the controller output value show the same trend, the reasonableness is positive; otherwise, the reasonableness is negative. When the system gain is negative, within a predetermined range, if the controller output value or the measured value remains unchanged, the reasonableness is zero; if the trend extraction results of the measured value and the trend extraction results of the controller output value show the same trend, the reasonableness is negative; otherwise, the reasonableness is positive.

[0092] It should be noted that the predetermined range here refers to a period before and after the current data point. For example, if the current data point is the tenth data point, the predetermined range can be from the fourth data point to the tenth data point. The trend of the fitting results of the measured values ​​and the fitting results of the controller output values ​​refers to the trend of the fitting results within this range.

[0093] Alternatively, please refer to Figure 1 In step 5, model identification is performed on the PID loop to obtain the loop model. Specifically, the least squares principle is used to identify the PID loop model, resulting in the following loop model: Where K represents the gain of the PID loop, T is the time constant, and τ represents the pure time delay.

[0094] Specifically, in this embodiment, the least squares principle is used to identify the PID loop model, resulting in the loop model as follows: Where K represents the gain of the PID loop, T is the time constant, τ represents the pure time delay, and s is the special symbol for the Laplace transform. The specific method of the least squares principle can be found in existing techniques and will not be elaborated here.

[0095] It should be noted that using the least squares principle for model identification is only one implementation method in this embodiment and is not intended to limit this application. In other embodiments, other methods can also be used for model identification, such as the maximum likelihood estimation principle. In practical use, a suitable model identification method can be selected according to needs, and this application does not limit it in this regard.

[0096] Optionally, refer to Figure 1In step 5, the PID loop parameters are tuned, specifically using the ZN method. Specifically, in this embodiment, the classic ZN (Ziegler-Nichols) empirical tuning method, also known in engineering as the fly-up curve method, is used to tune the PID loop parameters. Then, corresponding PID parameters are given according to different DCS models. For example, when the DCS model is a Supcon ECS700, if the PID parameters are selected in PI form, then... T i =3.3τ,T d =0; if the PID parameter is selected in PID form, then T i =2.0τ,T d =0.5τ. Where P represents the proportionality constant, and T i T represents the integration time. d Let T represent the derivative time, K represent the gain of the PID loop, T be the time constant, and τ represent the pure time delay.

[0097] It should be noted that using the ZN method to tune the PID loop parameters is only one implementation method in this embodiment and is not intended to limit this application. The ZN method is an empirical tuning method. In other embodiments, other empirical tuning methods (such as the Cohen-Coon method) or more complex methods such as robust tuning methods can be used to obtain the PID parameters. In practical use, a suitable parameter tuning method can be selected according to the specific needs, and this application does not limit it in this regard.

[0098] Based on the same inventive concept, this invention also provides a PID loop parameter intelligent tuning system based on key indicators. Figure 3 This is a schematic diagram of a tuning system provided in an embodiment of the present invention. Please refer to it. Figure 3 The present invention provides a PID loop parameter intelligent tuning system 100 based on key indicators, comprising:

[0099] The acquisition module 10 is used to acquire the operating parameters of the PID loop from the distributed control system (DCS).

[0100] The setting module 20 is used to set key indicators that reflect the control performance of the PID loop;

[0101] The first calculation module 30 is used to calculate the actual values ​​of key indicators based on the operating parameters of the PID loop.

[0102] Setting module 40 is used to set the threshold of key indicators according to the characteristics of the PID loop;

[0103] The judgment module 50 is used to determine whether the PID loop meets the control performance degradation requirement based on the relationship between the actual value and the threshold. If yes: a control performance degradation notification is sent, and the intelligent PID parameter tuning function is activated; data segments that reflect the characteristics of the PID loop and are suitable for modeling are selected from the operating parameters; the PID loop is modeled based on the data segments to obtain the loop model; the PID loop parameters are tuned according to the loop model and the type of distributed control system (DCS) to obtain suggested values ​​for the PID parameters; the suggested values ​​for the PID parameters are input into the DCS; the operating parameters of the PID loop are collected again, and the actual values ​​of key indicators are calculated; if no: the operating parameters of the PID loop are collected again, and the actual values ​​of key indicators are calculated.

[0104] For details, please refer to Figure 3 The PID loop parameter intelligent tuning system based on key indicators provided in this embodiment first uses the acquisition module 10 to acquire the operating parameters of the PID loop from the distributed control system (DCS). The operating parameters acquired here can be selected as needed, such as including but not limited to the setpoint SV, measured value PV, and controller output value OP. In addition, it also includes the proportional coefficient P, integral time I, and derivative time D of the PID controller.

[0105] After acquiring the operating parameters of the PID loop, the setting module 20 sets key indicators reflecting the control performance of the PID loop, and the first calculation module 30 calculates the actual values ​​of the key indicators based on the operating parameters of the PID loop. Here, key indicators can be, for example, the coefficient of variation and the accuracy coefficient. The coefficient of variation can evaluate the characteristics of the system recovering to steady state after deviating from steady state due to external disturbances, and the accuracy coefficient can evaluate the control accuracy characteristics between the measured value and the set value of the controlled variable during the process of the system reaching steady state.

[0106] Then, through the setting module 40, thresholds for key indicators are set according to the characteristics of the PID loop. The thresholds are used to determine whether the calculated actual values ​​reflect that the loop is maintaining an optimal operating state. The characteristics of the PID loop here can be, for example, the importance of the loop, interference, etc.

[0107] After obtaining the actual values ​​and thresholds of key indicators, the judgment module 50 compares the magnitudes of the actual values ​​and thresholds. Based on this relationship, it determines whether the PID loop meets the requirements for decreased control performance. If the PID loop meets these requirements, it indicates that the control loop is not operating in its optimal state, and a notification of decreased control performance is sent, activating the intelligent PID parameter tuning function. During intelligent parameter tuning, data segments that reflect the characteristics of the PID loop and are suitable for modeling are first selected from the collected operating parameters. Based on the selected data segments, the PID loop is modeled; for example, the least squares principle can be used for model identification to obtain the loop model.

[0108] After obtaining the loop model, the PID loop parameters are tuned according to the loop model and the type of Distributed Control System (DCS), yielding suggested values ​​for the PID parameters. The classic ZN empirical tuning method can be used for parameter tuning, and then corresponding PID parameters are given based on different DCS models. For example, when the DCS model is a Supcon ECS700, if the PID parameters are selected in PI form, then... T i =3.3τ,T d =0; if the PID parameter is selected in PID form, then T i =2.0τ,T d =0.5τ. Where P represents the proportionality constant, and T i T represents the integration time. d Let T represent the derivative time, K represent the gain of the PID loop, T be the time constant, and τ represent the pure time delay.

[0109] After obtaining the suggested values ​​of the PID parameters, input the suggested values ​​of the PID parameters into the distributed control system (DCS). Then, collect the operating parameters of the PID loop again and calculate the actual values ​​of key indicators. By repeating the above steps, periodically judge the control performance of the PID loop. When the control performance of the PID loop declines, timely intelligent parameter tuning can be performed to keep the control loop working in the best state.

[0110] When the PID loop fails to meet the requirements for degraded control performance, it indicates that the control loop is operating at its optimal state, and parameter tuning is unnecessary. Therefore, continue collecting the operating parameters of the PID loop and calculating the actual values ​​of key indicators. By repeating the above steps, periodically assess the control performance of the PID loop. When the control performance of the PID loop deteriorates or the control effect is poor, timely intelligent parameter tuning can be performed to ensure that the control loop always operates at its optimal state.

[0111] The intelligent PID loop parameter tuning system based on key indicators provided by this invention sets key indicators that reflect control performance. By calculating the actual values ​​of these key indicators and judging the control performance of the PID loop based on the relationship between the actual values ​​and thresholds, the system can intelligently tune parameters in a timely manner when the control performance of the PID loop deteriorates or the control effect is poor. This not only ensures that the control loop operates in an optimal state but also effectively reduces the operational intensity and difficulty for operators. Furthermore, by repeatedly collecting operating parameters and calculating the actual values ​​of key indicators, changes in loop characteristics caused by changes in production conditions such as load can be detected at any time. This allows for timely intelligent parameter tuning during shifts in operating states, further contributing to maintaining the control loop in an optimal state.

[0112] Optionally, the operating parameters of the PID loop include the setpoint, measured value, and controller output value within the set time period. Figure 4 For another structural schematic diagram of the tuning system provided in this embodiment of the invention, please refer to... Figure 4 The system also includes:

[0113] Extraction module 501 is used to extract trends from measured values ​​and controller output values ​​to obtain trend extraction results;

[0114] Analysis module 502 performs a reasonableness analysis on each data point based on the trend extraction results to obtain the reasonableness of each data point;

[0115] The segmentation module 503 is used to divide the set value, measured value and controller output value into multiple sliding data segments using a sliding window method;

[0116] The second calculation module 504 is used to obtain the average value of the rationality of all data points in the sliding data segment based on the rationality of each data point and the sliding data segment, which is denoted as the rationality of the sliding data segment.

[0117] The calibration module 505 is used to set the rationality of the sliding data segment to zero when the automatic state within the sliding data segment is automatic and the set value remains unchanged.

[0118] The selection module 506 is used to select sliding data segments with a reasonableness of the first value as suitable data segments for modeling.

[0119] For details, please refer to Figure 4The tuning system provided in this embodiment also includes an extraction module 501, an analysis module 502, a segmentation module 503, a second calculation module 504, a calibration module 505, and a selection module 506. When selecting data segments from the operating parameters that reflect the characteristics of the PID loop and are suitable for modeling, the extraction module 501 first performs trend extraction on the measured values ​​and the controller output values ​​to obtain the trend extraction results. During trend extraction, binomial fitting and dynamic programming can be used, selecting an appropriate number of segments and segmentation points, and using the corresponding binomial fitting results as the final trend extraction result.

[0120] After trend fitting, each data point corresponds to a reasonableness score. Therefore, after obtaining the trend extraction results, the analysis module 502 uses the trend extraction results to perform reasonableness analysis on each data point, thus obtaining the reasonableness score of each data point. Then, the segmentation module 503 uses a sliding window method to divide the setpoint, measured value, and controller output value into multiple sliding data segments. When using a sliding window for segmentation, the window width parameter can be determined according to the loop type. For example, for flow and pressure loops, the window width parameter is set to 133, while for temperature and level loops, the window width parameter is set to 800. Then, the segmentation points for window division are determined using the width parameter, thereby determining each sliding data segment.

[0121] After obtaining the reasonableness of the sliding data segments and each data point, the second calculation module 504 calculates the average reasonableness of all data points in each sliding data segment based on the reasonableness of each data point and the sliding data segment, denoted as the reasonableness of the sliding data segment. After obtaining the reasonableness of each sliding data segment, the module determines whether reasonableness correction is needed based on the automatic state and set value within the sliding data segment. If the automatic state within the sliding data segment is automatic and the set value remains unchanged, correction is required, and the correction module 505 sets the reasonableness of the corresponding sliding data segment to zero. Otherwise, no correction is needed, and the process proceeds directly to the next step.

[0122] After obtaining the reasonableness of each sliding data segment, the selection module 506 selects the sliding data segment with the highest reasonableness value as the suitable data segment for modeling. Here, the value with the highest reasonableness of the sliding data segment is taken as the first value.

[0123] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of one embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing the present invention.

[0124] Those skilled in the art will understand that the modules in the apparatus of the embodiments can be distributed in the apparatus of the embodiments as described in the embodiments, or they can be located in one or more devices different from this embodiment with corresponding changes. The modules of the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent tuning of PID loop parameters based on key indicators, characterized in that, include: The operating parameters of the PID loop are acquired from the distributed control system (DCS); the operating parameters of the PID loop include the set value, measured value and controller output value within a set time period. Define key indicators that reflect the control performance of the PID loop; Calculate the actual values ​​of the key indicators based on the operating parameters of the PID loop; Based on the characteristics of the PID loop, set the threshold values ​​for the key indicators; Based on the relationship between the actual value and the threshold, determine whether the PID loop meets the control performance degradation requirement. If so, send a control performance degradation notification and start the intelligent PID parameter tuning function; select data segments from the operating parameters that can reflect the characteristics of the PID loop and are suitable for modeling. The PID loop is modeled based on the data segment to obtain a loop model; the PID loop is then tuned according to the loop model and the type of the distributed control system (DCS) to obtain suggested values ​​for the PID parameters; these suggested values ​​are then input into the DCS; the operating parameters of the PID loop are continuously collected, and the actual values ​​of the key indicators are calculated; otherwise, the operating parameters of the PID loop are directly collected, and the actual values ​​of the key indicators are calculated. Specifically, the data segments selected from the operating parameters that reflect the characteristics of the PID loop and are suitable for modeling are as follows: Trend extraction is performed on the measured values ​​and the controller output values ​​to obtain trend extraction results; Based on the trend extraction results, a reasonableness analysis is performed on each data point to obtain the reasonableness of each data point; By using a sliding window method, the set value, the measured value, and the controller output value are divided into multiple sliding data segments; Based on the rationality of each data point and the sliding data segment, the average rationality of all data points in the sliding data segment is obtained and denoted as the rationality of the sliding data segment. When the automatic state within the sliding data segment is automatic and the set value remains unchanged, the rationality of the sliding data segment is set to zero. The sliding data segment whose reasonableness is a first value is taken as the data segment suitable for modeling.

2. The intelligent tuning method for PID loop parameters based on key indicators according to claim 1, characterized in that, The operating parameters of the PID loop include at least n measured values ​​PV within a set time period. i and n setpoints SV i The key indicators include the coefficient of variation (Cov) and the accuracy coefficient (Accu); the actual values ​​of the key indicators include the calculated values ​​of the coefficient of variation (Cov) and the calculated values ​​of the accuracy coefficient (Accu). Based on the operating parameters of the PID loop, the actual values ​​of the key indicators are calculated, specifically as follows: According to the formula Calculate the coefficient of variation Cov, where Ave represents the n measured values ​​PV. i The mean, where Std represents the standard deviation; the mean Ave is... The standard deviation Std is According to the formula Calculate the accuracy coefficient Accu.

3. The intelligent tuning method for PID loop parameters based on key indicators according to claim 2, characterized in that, The thresholds include the threshold for the coefficient of variation Cov and the threshold for the accuracy coefficient Accu. The determination of whether the PID loop meets the control performance degradation requirement is based on the relationship between the actual value and the threshold. Specifically: The calculated value of the coefficient of variation Cov and the threshold, and the calculated value of the accuracy coefficient Accu and the threshold are compared respectively. When the calculated value of the coefficient of variation Cov is greater than the threshold of the coefficient of variation Cov and the calculated value of the accuracy coefficient Accu is less than the threshold of the accuracy coefficient Accu, it indicates that the control performance degradation requirement is met. Otherwise, the requirement for reduced control performance will not be met.

4. The intelligent tuning method for PID loop parameters based on key indicators according to claim 3, characterized in that, The threshold for the coefficient of variation Cov is 2, and the threshold for the accuracy coefficient Accu is 0.

8.

5. The intelligent tuning method for PID loop parameters based on key indicators according to claim 1, characterized in that, Based on the trend extraction results, a reasonableness analysis is performed on each data point to obtain the reasonableness of each data point, specifically: When the system gain is positive, within a predetermined range, if the controller output value or the measured value remains unchanged, the reasonableness is zero; if the trend extraction result of the measured value and the trend extraction result of the controller output value change in the same direction, the reasonableness is positive; otherwise, the reasonableness is negative; when the system gain is negative, within a predetermined range, if the controller output value or the measured value remains unchanged, the reasonableness is zero. If the trend extraction result of the measured value and the trend extraction result of the controller output value have the same trend, then the reasonableness is negative; otherwise, the reasonableness is positive.

6. The intelligent tuning method for PID loop parameters based on key indicators according to claim 1, characterized in that, The PID loop is model identified to obtain the loop model, specifically: The PID loop was modeled using the least squares principle, resulting in the following loop model: Where K represents the gain of the PID loop, T is the time constant, τ represents the pure time delay, and s is the special symbol for the Laplace transform.

7. The intelligent tuning method for PID loop parameters based on key indicators according to claim 6, characterized in that, The parameters of the PID loop are tuned as follows: The ZN method is used to tune the parameters of the PID loop.

8. A PID loop parameter intelligent tuning system based on key indicators, characterized in that, include: The data acquisition module is used to acquire the operating parameters of the PID loop from the distributed control system (DCS); the operating parameters of the PID loop include the setpoint, measured value, and controller output value within a set time period. The setting module is used to set key indicators that reflect the control performance of the PID loop. The first calculation module is used to calculate the actual value of the key indicator based on the operating parameters of the PID loop. The setting module is used to set the threshold of the key indicators according to the characteristics of the PID loop; The judgment module is used to determine whether the PID loop meets the control performance degradation requirement based on the relationship between the actual value and the threshold. If so, it sends a control performance degradation notification, starts the intelligent tuning function of PID parameters, and filters data segments from the operating parameters that can reflect the characteristics of the PID loop and are suitable for modeling. The PID loop is modeled based on the data segment to obtain a loop model; the PID loop is then tuned according to the loop model and the type of the distributed control system (DCS) to obtain suggested values ​​for the PID parameters; these suggested values ​​are then input into the DCS; the operating parameters of the PID loop are continuously collected, and the actual values ​​of the key indicators are calculated; otherwise, the operating parameters of the PID loop are directly collected, and the actual values ​​of the key indicators are calculated. The system also includes: The extraction module is used to extract trends from the measured values ​​and the controller output values ​​to obtain trend extraction results; The analysis module performs a reasonableness analysis on each data point based on the trend extraction results to obtain the reasonableness of each data point; The segmentation module is used to divide the set value, the measured value, and the controller output value into multiple sliding data segments using a sliding window method. The second calculation module is used to obtain the average value of the rationality of all data points in the sliding data segment based on the rationality of each data point and the sliding data segment, and denoted as the rationality of the sliding data segment. The calibration module is used to set the rationality of the sliding data segment to zero when the automatic state within the sliding data segment is automatic and the set value remains unchanged. The selection module is used to select the sliding data segments whose reasonableness is a first value as suitable data segments for modeling.

Images