Control loop monitoring method based on expert experience and multivariate indexes

Abstract

The invention relates to a control loop monitoring method based on expert experience and multivariate indexes. The method comprises the following steps: acquiring online data of a control loop in real time; a control loop quality comprehensive index is calculated for the online real-time data, the control loop quality comprehensive index evaluates the control loop process from the aspects of reliability, deviation, timeliness and oscillation, and a control loop evaluation index coefficient is determined based on expert experience; and grade division is carried out based on the comprehensive performance index score of the current control loop, and real-time early warning is carried out on the control loop based on different grades. Compared with the prior art, by combining expert experience and diversified indexes, the performance of the analog quantity control loop is comprehensively reflected, the short-term sudden change influence caused by negligence of operators or control disturbance is effectively prevented, the real-time monitoring capability of the operators on the control process can be remarkably improved, and the control accuracy is improved. And the operation safety and stability of the unit are improved.

Description

Technical Field

[0001] The present invention relates to the field of control loops in the power generation industry, and in particular to a control loop monitoring method based on expert experience and multivariate indicators. Background Art

[0002] With the continuous increase in residents' electricity demand and the expansion of the scale of new energy grid connection in the power grid, the frequency of grid fluctuations has increased significantly, forcing the power generation field to make more flexible adjustments. Therefore, the requirements for the response amplitude and frequency of units in the power generation field are becoming increasingly stringent. Taking traditional thermal power units as an example, during the load change process, traditional thermal power units need to monitor the operating conditions of the units at all times. However, the traditional monitoring method only relies on the operator to observe the response status of the unit with the naked eye, which is not enough to meet the current complex operating environment and the requirements of efficient operation.

[0003] In the power generation process, especially for traditional thermal power units, there are many observation parameters, high degree of control loop coupling, and complex cross-systems. The alarm for observed variables is generally judged by quality, high and low limit, and three-way judgment. When the control deviation is too large, the control loop is directly switched to manual mode, which is relatively simple and cannot achieve real-time monitoring of the control process performance. This limitation often leads to subsequent larger deviations or shocks, and the related control loops are automatically switched to manual mode, which reduces the overall control effect of the unit, affects the output and safety, and even causes serious grid-tripping accidents.

[0004] In the field of power generation, traditional early warning methods mainly rely on single-point monitoring, feedback and command deviation monitoring. Among them, single-point monitoring usually provides early warning through quality judgment, high and low limit judgment, and three-way judgment. In control loop monitoring, it is usually necessary to directly switch to manual intervention when the deviation between command and feedback exceeds the fixed limit, or to calculate overshoot, rise time, response time and other indicators through disturbance tests for post-test evaluation. These indicators often exist independently, making it difficult to intuitively monitor the overall status of the control loop, and there may be a certain lag, which affects the effectiveness monitoring of real-time control.

[0005] At present, the evaluation of control loops is usually carried out through comprehensive judgment of multiple indicators, but the selection of multiple indicators is not comprehensive, especially when a single-point instantaneous disturbance occurs and quickly returns to normal, the evaluation results are difficult to maintain their accuracy. In this case, it is also necessary to query the historical data trend to identify the cause, otherwise effective troubleshooting cannot be carried out. In addition, the evaluation method is relatively rigid, and a unified coefficient is usually used to synthesize the evaluation index, ignoring the inconsistency of the control process characteristics of each loop. Summary of the invention

[0006] The purpose of the present invention is to overcome the complexity and redundancy problems of the above-mentioned existing power generation intelligent control. Due to the limited monitoring ability, it is difficult for operating personnel to perceive abnormal changes in the control loop in time, which leads to processing delays and brings potential safety hazards to the stable operation of the unit. A control loop evaluation and monitoring method based on expert experience and multi-indicators is provided.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A control loop monitoring method based on expert experience and multivariate indicators, comprising the following steps:

[0009] Acquire online real-time data of the control loop, and calculate the comprehensive quality index of the furnace analog control loop based on the online real-time data, wherein the comprehensive quality index of the control loop evaluates the control loop process from the perspectives of reliability, deviation, timeliness and oscillation;

[0010] Experts in professional fields determine the weight coefficients of multiple indicators based on the characteristics of the control loop;

[0011] The control loop is graded based on the comprehensive performance index score of the current control loop, and real-time warning is given to the control loop based on different levels.

[0012] As a control loop evaluation monitoring technical solution, the control loop evaluation index coefficient is determined based on expert experience, as follows:

[0013] Experts divided by professional fields assign values ​​to the importance of each control loop evaluation index to the control loop, and construct the expert experience matrix E one by one;

[0014] The expert experience matrix E is characterized to obtain the feature vector w, and a consistency test is performed to determine the validity of the feature vector;

[0015] The expert feature vectors that pass the validity test are averaged and normalized to obtain the weight vector W, which is the loop control evaluation index coefficient.

[0016] As a control loop evaluation technical monitoring scheme, the analog control loop quality comprehensive index S includes a reliability index S1, a deviation index, a timeliness index S4 and an oscillation index, wherein the deviation index includes a real-time deviation index S2 and a maximum deviation index S3; the oscillation index includes a control command oscillation index S5 and a controlled variable oscillation index S6.

[0017] As a control loop evaluation and monitoring technical solution, the reliability index S1 is calculated as follows: judging according to the forced or locked state of the control loop, if there is a forced or locked state, the reliability index S1=0, otherwise S1=100.

[0018] As a control loop evaluation and monitoring technical solution, the real-time deviation index S2 is calculated as follows: Calculate the absolute value of the real-time control deviation a dev , set the deviation dead zone a dead and the upper limit of deviation a uplim ;

[0019] If the absolute value of the real-time control deviation is a dev In the dead zone dead Real-time deviation index within the range S2 = 100;

[0020] If the absolute value of the real-time control deviation is a dev Exceeding the upper limit uplim , then the real-time deviation index S2 = 0;

[0021] If the absolute value of the real-time control deviation is a dev Satisfy [a dead ,a uplim ], and the intermediate interpolation method is used to evaluate the real-time deviation index S2.

[0022] As a control loop evaluation monitoring technical solution, the maximum deviation index S3 is calculated as follows: Using a dynamic sliding window, calculate the current sliding window T g1 The absolute value of the maximum deviation within a dm ;

[0023] If the current sliding window T g1 The absolute value of the maximum deviation within a dm Exceeding the maximum loop deviation limit a dm,lim When S3=0, otherwise S3=100.

[0024] As a control loop evaluation and monitoring technical solution, the timeliness index S4 is calculated as follows: When the control command changes beyond the dead zone, the controlled quantity responds to the control command change and the response direction is consistent with the control command. r The value is the difference between the moment when the current control instruction changes and the moment when the controlled quantity meets the response requirement;

[0025] If the controlled variable response time Tr is less than the optimal response time T fine , then S4=100;

[0026] If the response time Tr of the controlled variable is greater than the tolerable time T bear , then S4=0;

[0027] If the controlled variable response time Tr is in T fine and T bear If the value is less than S4, the interpolation method is used to evaluate S4.

[0028] As a control loop evaluation monitoring technology solution, the control command oscillation index S5 is calculated as follows:

[0029] Using sliding time window T g2 Perform extreme value processing on the control instruction timing array in the current window to form a maximum value vector and a minimum value vector;

[0030] Perform secondary extremization on the maximum value vector and take its maximum value vector as R up,o , perform secondary extremization on the minimum value vector and take its minimum value vector as R dn,o；

[0031] Calculate the number of control command oscillations n shark,o , and round down;

[0032] n shark,o =(len(R up,o )+len(R dn,o )) / 2

[0033] If the control command oscillates times n shark,o Greater than the maximum tolerable number of control command oscillations n bear,o , then S5 = 0; if the control command oscillates times n shark,o Less than or equal to the maximum tolerable number of control command oscillations n bear,o , then S5=100.

[0034] As a control loop evaluation and monitoring technical solution, the controlled variable feedback oscillation index S6 is calculated as follows:

[0035] Using sliding time window T g2 The controlled variable time series array in the current window is subjected to extreme value processing to form a maximum value vector and a minimum value vector; the sliding time window of the controlled variable feedback oscillation indicator is consistent with the sliding time window of the control instruction oscillation indicator;

[0036] Perform secondary extremization on the maximum value vector and take its maximum value vector as R up,f , perform secondary extremization on the minimum value vector and take its minimum value vector as R dn,f ;

[0037] Calculate the number of oscillations n of the controlled quantity shark,f , and round down;

[0038] n shark,f =(len(R up,f )+len(R dn,f )) / 2

[0039] If the regulated quantity oscillates times n shark,f Greater than the maximum tolerable number of feedback oscillations of the controlled variable n bear,f, then S6=0; if the regulated quantity oscillates n times shark,f Less than or equal to the maximum tolerable number of feedback oscillations of the controlled quantity n bear,f , then S6=100.

[0040] As a control loop evaluation and monitoring technical solution, the control loop is graded based on the comprehensive performance index score of the current control loop, and a real-time early warning is given to the control loop based on different grades, as follows:

[0041] When the comprehensive performance index of the current control loop is greater than or equal to the first threshold, it means that the current loop control is excellent and no manual intervention is required;

[0042] When the comprehensive performance index of the current control loop is greater than or equal to the second threshold and less than the first threshold, it means that there is a certain disturbance in the current loop control, reminding the monitoring personnel to always pay attention to the changes of relevant parameters in the control loop;

[0043] When the comprehensive performance index of the current control loop is greater than or equal to the third threshold and less than the second threshold, it means that the current loop control deviation is seriously beyond the line, the actuator is stuck, the command feedback is oscillating, and other serious situations occur. It is necessary to closely monitor the current control loop and arrange personnel to investigate the relevant causes;

[0044] When the comprehensive performance index of the current control loop is less than the third threshold, it means that the performance of the current control loop is too poor, and under the setting, the control loop is switched to manual control.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] The control loop evaluation and monitoring method based on expert experience and multivariate indicators proposed in the present invention combines expert experience and multivariate indicators. Experts in professional fields such as power plant operation and control determine the weight coefficients of multivariate indicators based on the control loop characteristics. Multivariate indicators evaluate the control loop process from multiple angles such as reliability, deviation, timeliness and oscillation, and establish a comprehensive control loop quality index. It can fully reflect the response characteristics of the current control loop and the performance of the analog control loop, and is particularly suitable for monitoring important control loops in power plants. It enables operating personnel to quickly intervene in the control process, thereby improving the efficiency and stability of the power generation system.

[0047] And through this method, it is possible to achieve early warning of the current control quality and record the control effect over a period of time. This process effectively prevents short-term mutations caused by operator negligence or control disturbances, thereby improving the monitoring of the control loop. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a flow chart of the control loop monitoring method based on expert experience and multivariate indicators of the present invention;

[0049] Figure 2 This is a diagram showing the evaluation results of the furnace negative pressure control loop in one of the specific implementation examples of the present invention. DETAILED DESCRIPTION

[0050] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0051] Example 1

[0052] Aiming at the online monitoring and early warning of control loops in the power generation field, this patent proposes a control loop evaluation and monitoring method based on expert experience and multi-indicators. Experts in professional fields such as power plant operation and control determine the weight coefficients of multi-indicators based on the characteristics of the control loop. Multi-indicators evaluate the control loop process from multiple angles such as reliability, deviation, timeliness and oscillation, and establish comprehensive control loop quality indicators. The specific steps are as follows: Figure 1 As shown:

[0053] Obtain online real-time data of the control loop, pre-process the acquired data and score it based on various multivariate indicators;

[0054] Determine the control loop evaluation index coefficient based on expert experience: divide experts by professional field, and have each expert assign a value to the importance of the control loop evaluation index to the control loop; summarize the expert importance assignment feature vectors that have passed the consistency test to obtain the control loop evaluation index coefficient;

[0055] Combine the sub-index scoring results with the control loop evaluation index coefficient to score the comprehensive performance index of the control loop;

[0056] The control loop is graded based on the comprehensive performance index score of the current control loop, and real-time warning is given to the control loop based on different levels.

[0057] Among them, the comprehensive index S of control loop quality is composed of reliability index S1, deviation index (real-time deviation index S2 and maximum deviation index S3), timeliness index S4 and oscillation index (control command oscillation index S5 and controlled variable oscillation index S6) and other indicators.

[0058] The control loop evaluation index coefficients are determined based on expert experience, and a 6*6-dimensional expert experience matrix E is constructed based on the number of evaluation indexes, as shown below. Experts divided by professional fields assign values ​​to the importance of each control loop evaluation index to the control loop, where x ij is the comparison of the importance of index i and index j, and the importance relationship is x ij =1 / x jiThe expert experience matrix is ​​characterized to obtain the feature vector w = [w1, w2, ..., w6], and a consistency test is performed to determine the validity of the feature vector. The m-bit expert feature vectors that pass the validity test are averaged and normalized to obtain the weight vector W = [W1, W2, ..., W6], which is the loop control evaluation index coefficient. * Corresponding evaluation index S * The coefficient of .

[0059]

[0060] Reliability index S1 calculation: judge according to the forced or locked state of the control loop. If there is a forced or locked state, the reliability index S1=0, otherwise S1=100.

[0061] Calculation of real-time deviation index S2: The absolute value of real-time control deviation (deviation between set value and measured value) is a dev , and set the deviation dead zone a dead and the upper limit of deviation a uplim In the dead zone a dead In the range S2 = 100; if a dev Exceeding the upper limit uplim , then S2=0; if a dev Satisfy dead, a uplim ] and the intermediate interpolation method was used to evaluate S2.

[0062] Calculation of the maximum deviation index S3: Using dynamic sliding window T g1 Calculate the current T g1 The absolute value of the maximum deviation within is a dm , if a dm Exceeding the maximum loop deviation limit a dm,lim When S3=0, otherwise S3=100.

[0063] Calculation of timeliness index S4: Based on the response time T of the controlled quantity r When the control command changes beyond the dead zone, the controlled variable responds to the control command change and the response direction is consistent with the control command, the response time T r The value is the difference between the time when the current control command changes and the time when the controlled quantity meets the response requirements. fine , then S4=100; Tr is greater than the tolerable time T bear , then S4=0; Tr is in T fine and T bear If the value is less than S4, the interpolation method is used to evaluate S4.

[0064] Calculation of the control command oscillator S5: Using a sliding time window Tg2 The control instruction timing array {o1,o2,...,o tg2} is subjected to extremization processing to form a maximum vector and a minimum vector. The maximum vector is subjected to secondary extremization and its maximum vector is taken as R up,o , the minimum value vector is secondary extremized and its minimum value vector is R dn,o . Oscillation number n shark,o Calculate as follows and round down. shark,o Greater than the maximum tolerable number of control command oscillations n bear,o The score is 0 if this is the case, and 100 if this is not the case.

[0065] n shark,o =(len(R up,o )+len(R dn,o )) / 2

[0066] Calculation of the regulated quantity feedback oscillator index S6: Its sliding time window and the control command oscillator index sliding time window T g2 Keep consistent, for the controlled variable timing array {f1,f2,...,f tg2} is subjected to extremization processing to form a maximum vector and a minimum vector. The maximum vector is subjected to secondary extremization and its maximum vector is taken as R up,f , the minimum value vector is secondary extremized and its minimum value vector is R dn,f . Oscillation number n shark,f Calculate as follows and round down. shark,f Greater than the maximum tolerable number of feedback oscillations of the controlled variable n bear,f The score is 0 if this is the case, and 100 if this is not the case.

[0067] n shark,f =(len(R up,f )+len(R dn,f )) / 2

[0068] Calculate the comprehensive index of control loop quality S = S1*W1+S2*W2+…+S6*W6, S∈[0,100], and set the weight coefficient of each index corresponding to W1 to W6 respectively.

[0069] According to the above calculation, the comprehensive performance index S of the current control loop is divided into grades Ⅰ, Ⅱ, Ⅲ, and Ⅳ: when S is greater than or equal to 95, the evaluation grade is Ⅰ, which means that the current loop control is excellent and no manual intervention is required; when S is greater than or equal to 85 and less than 95, the control loop performance is Ⅱ, which means that there is a certain disturbance in the current loop control, and the monitoring personnel need to pay attention to the changes of relevant parameters in the control loop; when S is greater than or equal to 75 and less than 85, the control loop performance is Ⅲ, which means that the current loop control deviation is seriously over the line, the actuator is stuck, and the command feedback is oscillating. Serious situations such as the current control loop needs to be closely monitored and personnel are arranged to investigate related causes. When S is less than 75, the control loop performance is Ⅳ, which means that the current control loop performance is too poor. If necessary, the control loop can be switched to manual control.

[0070] The solution of the present invention comprehensively reflects the performance of analog control loops by combining expert experience and diversified indicators, and is particularly suitable for monitoring important control loops in power plants. Through this method, early warning of the current control quality can be achieved, and the control effect over a period of time can be recorded. This process effectively prevents short-term mutations caused by operator negligence or control disturbances, thereby improving the monitoring of control loops.

[0071] Example 2

[0072] In this embodiment, a specific implementation example of the above-mentioned scheme of the present invention is provided. The furnace negative pressure control loop of a million-kilowatt thermal power unit is selected for control loop evaluation implementation. The furnace negative pressure control loop is prone to feedback oscillation, large deviation and other adjustment processes, and the effectiveness of the method of the present invention is tested.

[0073] The comprehensive quality index S of the furnace negative pressure control loop is composed of reliability index S1, deviation index (real-time deviation index S2 and maximum deviation index S3), timeliness index S4 and oscillation index (control command oscillation index S5 and oscillation index S6).

[0074] Based on expert experience, the control loop evaluation index coefficients are determined, and a 6*6-dimensional expert experience matrix E is constructed from the number of evaluation indicators, as shown below. The validity of the feature vector is determined by characterization and consistency testing, and the weight vector W is obtained. t =[0.06587, 0.4556, 0.2111, 0.0668, 0.0441, 0.1563], which is the loop control evaluation index coefficient. In order to be more helpful for scoring and judging each index, the weight coefficient is retained to two decimal places, and the final weight coefficient W is obtained by fine-tuning W = [0.07, 0.45, 0.2, 0.07, 0.05, 0.16].

[0075]

[0076] Reliability index S1 calculation: judge according to the forced or locked state of the control loop. If there is a forced or locked state, the reliability index S1=0, otherwise S1=100.

[0077] Calculation of furnace negative pressure real-time deviation index S2: The absolute value of the deviation between the furnace negative pressure feedback and the set value is taken as the real-time deviation a dev , and set the deviation dead zone a dead is 40Pa, the upper limit of deviation is a uplim is 90Pa. dev In the dead zone dead In the range S2 = 100; if a dev Exceeding the upper limit uplim , then S2=0; if a dev Satisfy dead, a uplim ] and the intermediate interpolation method was used to evaluate S2.

[0078] Calculation of the maximum deviation index S3 of the furnace negative pressure: T g1 Set to 180s, a dm,lim Set to 100Pa, every time the controlled variable crosses the set value dead zone, calculate the current T g1 The maximum absolute value of the deviation in the time window is a dm , if a dm More than a dm,lim When S3=0, otherwise S3=100.

[0079] Calculation of furnace negative pressure response timeliness index S4: Based on the response time T of the controlled variable r OK, T fine Set to 5s, T bear Set to 10s. When the furnace negative pressure control command changes beyond the dead zone, the controlled variable responds to the control command change and the response direction is consistent with the control command, T r The value is the difference between the moment when the current control instruction changes and the moment when the controlled quantity meets the response requirement. r Less than T fine , then S4=100; T r Greater than T bear , then S4=0; T r In T fine and T bear If the value is less than S4, the interpolation method is used to evaluate S4.

[0080] Calculation of furnace negative pressure control command oscillation index S5: Sliding time window T g2 Set to 60s, the control instruction timing array {o1,o2,...,o tg2} is subjected to extremization processing to form a maximum vector and a minimum vector. The maximum vector is subjected to secondary extremization and its maximum vector is taken as R up,o , the minimum value vector is secondary extremized and its minimum value vector is R dn,o . Oscillation number n shark,o Calculate as follows and round down. bear,o Set to 3 times, if n shark,o Greater than n bear,o The score is 0 if this is the case, and 100 if this is not the case.

[0081] n shark,o =(len(R up,o )+len(R dn,o )) / 2

[0082] Calculation of the regulated quantity feedback oscillator index S6: Its sliding time window and the control command oscillator index sliding time window T g2 Keep consistent, for the time series array {f1,f2,...,f tg2} is subjected to extremization processing to form a maximum vector and a minimum vector. The maximum vector is subjected to secondary extremization and its maximum vector is taken as R up,f , the minimum value vector is secondary extremized and its minimum value vector is R dn,f . Oscillation number n shark,f Calculate as follows and round down. bear,f Set to 3 times, if n shark,f Greater than n bear,f The score is 0 if this is the case, and 100 if this is not the case.

[0083] n shark,f =(len(R up,f )+len(R dn,f )) / 2

[0084] The comprehensive quality index of the furnace negative pressure control loop S = S1*W1+S2*W2+…+S6*W6, and the weight coefficient of each index is set corresponding to W1 to W6 respectively.

[0085] According to the above calculation, the comprehensive performance index S of the furnace negative pressure control loop is divided into grades Ⅰ, Ⅱ, Ⅲ, and Ⅳ: when S is greater than or equal to 95, the evaluation grade is Ⅰ, which means that the current loop control is excellent and no manual intervention is required; when S is greater than or equal to 85 and less than 95, the control loop performance is Ⅱ, which means that there are certain disturbances in the furnace negative pressure loop control, and the monitoring personnel need to pay attention to the changes of relevant parameters in the control loop; when S is greater than or equal to 75 and less than 85, the control loop performance is Ⅲ, which means that the furnace negative pressure loop control deviation is seriously beyond the line, the actuator is stuck, and the command feedback is oscillating. Serious situations such as the current control loop needs to be closely monitored and personnel are arranged to investigate related causes. When S is less than 75, the control loop performance is Ⅳ, which means that the performance of the furnace negative pressure control loop is too poor. If necessary, the control loop can be switched to manual control.

[0086] like Figure 2 As shown, by verifying the real data of the furnace negative pressure control loop on a certain day, the furnace negative pressure instruction O is always maintained at -100Pa. In the first half of the time, the furnace negative pressure F showed the characteristics of high-frequency oscillation and large fluctuations, resulting in significant changes in indicators S2 and S6. Since indicators S2 and S6 account for a relatively large proportion, the comprehensive indicator S performs poorly, requiring operators to pay close attention to the current adjustment process. Although from the overall trend of the furnace negative pressure, the area is generally stable and the S indicator tends to be higher, there is still a problem of real-time deviation crossing the line. The evaluation indicators can be tracked and fed back in real time, thereby verifying the effectiveness of the method of the present invention.

[0087] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A control loop monitoring method based on expert experience and multivariate indicators, characterized in that the steps include: Acquire online real-time data of the control loop, and calculate the analog control loop quality comprehensive index based on the online real-time data, wherein the control loop quality comprehensive index evaluates the control loop process from the perspectives of reliability, deviation, timeliness and oscillation; Experts in professional fields determine the weight coefficients of multiple indicators based on the characteristics of the control loop; The control loop is graded based on the comprehensive performance index score of the current control loop, and real-time warning is given to the control loop based on different levels.

2. A control loop monitoring method based on expert experience and multivariate indicators according to claim 1, characterized in that: The control loop evaluation index coefficient is determined based on expert experience, as follows: Experts divided by professional fields assign values ​​to the importance of each control loop evaluation index to the control loop, and construct the expert experience matrix E one by one; The expert experience matrix E is characterized to obtain the feature vector w, and a consistency test is performed to determine the validity of the feature vector; The expert feature vectors that pass the validity test are averaged and normalized to obtain the weight vector W, which is the loop control evaluation index coefficient.

3. The control loop monitoring method based on expert experience and multivariate indicators according to claim 1 is characterized in that: The analog control loop quality comprehensive index S includes a reliability index S1, a deviation index, a timeliness index S4 and an oscillation index, wherein the deviation index includes a real-time deviation index S2 and a maximum deviation index S3; the oscillation index includes a control command oscillation index S5 and a controlled quantity oscillation index S6.

4. A control loop monitoring method based on expert experience and multivariate indicators according to claim 3, characterized in that: The reliability index S1 is calculated as follows: the reliability index S1 is calculated based on the forced or locked state of the control loop. If there is a forced or locked state, the reliability index S1=0, otherwise S1=100.

5. The control loop monitoring method based on expert experience and multivariate indicators according to claim 3 is characterized in that: The real-time deviation index S2 is calculated as follows: Calculate the absolute value of the real-time control deviation a dev , set the deviation dead zone a dead and the upper limit of deviation a uplim ; If the absolute value of the real-time control deviation is a dev In the dead zone dead The real-time deviation index within the range S2 = 100; If the absolute value of the real-time control deviation is a dev Exceeding the upper limit uplim , then the real-time deviation index S2 = 0; If the absolute value of the real-time control deviation is a dev Satisfy [a dead ,a uplim ], and the intermediate interpolation method is used to evaluate the real-time deviation index S2.

6. The control loop monitoring method based on expert experience and multivariate indicators according to claim 3 is characterized in that: The maximum deviation index S3 is calculated as follows: Using a dynamic sliding window, calculate the current sliding window T g1 The absolute value of the maximum deviation within dm ; If the current sliding window T g1 The absolute value of the maximum deviation within dm Exceeding the maximum loop deviation limit a dm,lim When S3=0, otherwise S3=100.

7. The control loop monitoring method based on expert experience and multivariate indicators according to claim 3 is characterized in that: The timeliness index S4 is calculated as follows: When the control command changes beyond the dead zone, the controlled quantity responds to the control command change and the response direction is consistent with the control command, the controlled quantity response time T r The value is the difference between the moment when the current control instruction changes and the moment when the controlled quantity meets the response requirement; If the controlled variable response time Tr is less than the optimal response time T fine , then S4=100; If the response time Tr of the controlled variable is greater than the tolerable time T bear , then S4=0; If the controlled variable response time Tr is in T fine and T bear If the value is less than S4, the interpolation method is used to evaluate S4.

8. The control loop monitoring method based on expert experience and multivariate indicators according to claim 3 is characterized in that: The control command oscillation index S5 is calculated as follows: Using sliding time window T g2 Perform extreme value processing on the control instruction timing array in the current window to form a maximum value vector and a minimum value vector; Perform secondary extremization on the maximum value vector and take its maximum value vector as R up,o , perform secondary extremization on the minimum value vector and take its minimum value vector as R dn,o； Calculate the number of control command oscillations n shark,o , and round down; n shark,o =(len(R up,o )+len(R dn,o )) / 2 If the control command oscillates times n shark,o Greater than the maximum tolerable number of control command oscillations n bear,o , then S5=0; if the control command oscillates times n shark,o Less than or equal to the maximum tolerable number of control command oscillations n bear,o , then S5=100.

9. A control loop monitoring method based on expert experience and multivariate indicators according to claim 8, characterized in that: The regulated quantity feedback oscillation index S6 is calculated as follows: Using sliding time window T g2 The controlled variable time series array in the current window is subjected to extreme value processing to form a maximum value vector and a minimum value vector; the sliding time window of the controlled variable feedback oscillation indicator is consistent with the sliding time window of the control instruction oscillation indicator; Perform secondary extremization on the maximum value vector and take its maximum value vector as R up,f , perform secondary extremization on the minimum value vector and take its minimum value vector as R dn,f ; Calculate the number of oscillations n of the controlled quantity shark,f , and round down; n shark,f =(len(R up,f )+len(R dn,f )) / 2 If the regulated quantity oscillates times n shark,f Greater than the maximum tolerable number of feedback oscillations of the controlled variable n bear,f , then S6=0; if the regulated quantity oscillates n times shark,f Less than or equal to the maximum tolerable number of feedback oscillations of the controlled quantity n bear,f , then S6=100.

10. The control loop monitoring method based on expert experience and multivariate indicators according to claim 1, characterized in that: The control loop is graded based on the comprehensive performance index score of the current control loop, and a real-time warning is given to the control loop based on different grades, as follows: When the comprehensive performance index of the current control loop is greater than or equal to the first threshold, it means that the current loop control is excellent and no manual intervention is required; When the comprehensive performance index of the current control loop is greater than or equal to the second threshold and less than the first threshold, it means that there is a certain disturbance in the current loop control, reminding the monitoring personnel to always pay attention to the changes of relevant parameters in the control loop; When the comprehensive performance index of the current control loop is greater than or equal to the third threshold and less than the second threshold, it means that the current loop control deviation is seriously beyond the line, the actuator is stuck, the command feedback is oscillating, and other serious situations occur. Closely monitor the current control loop and arrange personnel to investigate the relevant causes; When the comprehensive performance index of the current control loop is less than the third threshold, it means that the performance of the current control loop is too poor, and under the setting, the control loop is switched to manual control.

Images