Intelligent electric control valve operation optimization control method

Abstract

The invention discloses an intelligent electric control valve operation optimization control method. The method comprises the steps that actual process parameters and the valve opening degree are collected, and the discharging flow is predicted; constructing a sample library, and reasoning and judging historical data according to expert experience knowledge; dividing working condition evaluation grades by using a cloud model of cloud entropy optimization, and integrating subjective weights and objective weights by combining a game theory and an expert scoring matrix to obtain a final evaluation result; if the current working condition does not conform to the expected working condition, searching data most similar to the expected working condition from a sample library by utilizing a c-means cloud clustering intelligent search strategy, and outputting a valve opening value corresponding to the data as an optimized operation mode; and performing optimization control on the valve electric actuating mechanism through a multi-rate model prediction intelligent control algorithm. According to the method, full-period self-adaptive optimization control from flow prediction, operation state evaluation and intelligent optimization of operation parameters to model prediction control is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent electric-controlled valve operation status evaluation and optimization control, and in particular to an intelligent electric-controlled valve operation optimization control method. Background Art

[0002] In the context of the rapid development of industrial automation and intelligent manufacturing, the demand for efficient and accurate control systems is growing. As the core regulating component in the industrial process, the operating status of the valve is directly related to production efficiency, resource consumption and process safety. The precise adjustment of the valve opening is crucial for controlling key parameters such as flow and pressure. However, traditional valve control methods mostly rely on empirical settings or basic automation logic, and fail to make full use of the large amount of collected process data for in-depth analysis to optimize the control strategy. Existing systems have limitations in real-time monitoring and prediction of industrial process parameters and valve opening, and are often unable to keep up with the rapid changes in the industrial environment due to insufficient accuracy and slow response. In addition, the current working condition evaluation methods either rely too much on subjective judgment or rely solely on objective data, failing to achieve a comprehensive consideration of subjective and objective factors. The complexity and ever-changing needs of industrial systems require control systems to have higher adaptability and flexibility, while traditional systems are often unable to cope with complex or unknown working conditions, making it difficult to achieve precise control of valve opening and meet the response requirements of rapidly changing working conditions. Summary of the invention

[0003] In order to overcome the deficiencies in the above-mentioned prior art, the present invention provides an intelligent electronically controlled valve operation optimization control method. The method comprehensively utilizes a flow prediction model, a comprehensive operating condition evaluation model, an expert scoring decision model, an operating parameter intelligent optimization algorithm, a multi-rate model predictive control method, and constructs an expert knowledge base. By integrating artificial intelligence algorithms with expert experience, the present invention realizes full-cycle adaptive optimization control from flow prediction, operating status evaluation, intelligent optimization of operating parameters, to model predictive control. The core advantage of the present invention lies in its high adaptability and intelligence level. The flow prediction model can accurately predict the discharge flow rate and provide data support for subsequent control; the comprehensive operating condition evaluation model comprehensively evaluates the current operating status; the expert scoring decision model combines expert experience to provide decision support; the operating parameter intelligent optimization algorithm dynamically adjusts the valve opening to adapt to different operating conditions; the multi-rate model predictive control method ensures the accuracy and response speed of the valve action; and the expert knowledge base provides the system with rich reference information and decision-making basis. In order to achieve the above-mentioned technical objectives, the present invention adopts the following technical solutions:

[0004] An intelligent electronically controlled valve operation optimization control method comprises the following steps:

[0005] S1: Collect the actual process parameters and valve opening data in the industrial process, and use the lightweight block incremental geometry construction network algorithm to predict the discharge flow rate;

[0006] S2: Build a sample library based on historical parameter data, and use expert experience and knowledge to infer and judge historical data;

[0007] S3: Use the cloud model optimized by cloud entropy to classify the working condition evaluation level, combine game theory and expert scoring matrix to integrate subjective weight and objective weight to obtain the final evaluation result, and judge whether the current working condition meets the expected working condition;

[0008] S4: If the current working condition does not meet the expected working condition, a c-means cloud clustering intelligent search strategy is used to search for the data most similar to the expected working condition from the sample library, and the corresponding valve opening value is output as the optimized operation mode;

[0009] S5: Through the multi-rate model predictive intelligent control algorithm, the valve electric actuator is optimized to achieve precise control of the valve opening.

[0010] Compared with the prior art, the present invention has the following beneficial effects:

[0011] 1. The present invention proposes an intelligent electronically controlled valve operation optimization control method, which integrates a series of decision modules to form a comprehensive control process. This process covers key links such as flow prediction, operation status evaluation, intelligent optimization of operating parameters, and multi-rate model predictive control, aiming to solve the control problems caused by variable working conditions in industrial production. In particular, it can cope with those situations where it is difficult to achieve adaptive adjustment of valve opening to meet the expected working conditions through traditional methods.

[0012] 2. The present invention adopts a lightweight block incremental geometric construction network algorithm, which significantly improves the response speed of the system while ensuring the accuracy of discharge flow prediction. Its block incremental network structure enables the model to be updated online in real time, thus providing a fast and accurate flow prediction solution for industrial processes.

[0013] 3. The present invention divides the working condition evaluation levels through the cloud model optimized by cloud entropy, and uses game theory and expert scoring matrix to integrate subjective weights and objective weights to obtain the final evaluation results, giving full play to the advantages of expert experience while taking into account objective reflection data to obtain more reasonable indicator weights.

[0014] 4. The present invention uses intelligent search strategies, combined with c-means cloud clustering algorithm and correlation coefficient calculation method, to accurately screen out data highly similar to expected working conditions from the expert knowledge base. By using the valve opening corresponding to these data as the output of the control mode, the search sample space is effectively reduced, the search process is optimized, and the search efficiency and accuracy are greatly improved.

[0015] 5. The present invention designs a multi-rate model predictive control algorithm that can respond to changes in system status in real time, quickly adjust the control strategy, and make adjustments in advance by predicting the system status in the future, thereby avoiding possible deviations and achieving more precise control effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions of 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 creative work.

[0017] Figure 1 is a flow chart of the method of the present invention;

[0018] Figure 2 It is a network structure diagram of lightweight block incremental geometry construction in the present invention;

[0019] Figure 3 A schematic diagram of cloud model hierarchy constructed by optimizing cloud entropy in the present invention. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] Example

[0022] like Figure 1 As shown, a method for optimizing the operation of an intelligent electronically controlled valve comprises the following steps:

[0023] S1: Collect the actual process parameters and valve opening data in the industrial process, and use the lightweight block incremental geometry network algorithm to predict the discharge flow rate. The network structure is as follows: Figure 2 As shown; step S1 is as follows:

[0024] A geometrically constructed network with block increments is used as the prediction model, based on the collected historical process parameters, including flow rate, pressure, temperature, and the corresponding valve opening and discharge volume data, to obtain input data for training the neural network model. And output discharge flow data Y = {y1, y2, ..., y N},in is the process parameter data collected at time N, i is the number of process parameters, y N is the corresponding discharge flow data; all data are normalized according to the following formula to obtain

[0025]

[0026] The specific steps for building a prediction model are as follows:

[0027] S11: Initialize the maximum number of hidden layer nodes L max , maximum number of configuration times T max , critical tolerance error ε, random parameter range [-λ max ,λ max ] and the variation step size Δλ;

[0028] S12: Randomly configure T within the adaptive interval max The input weight ω of the Lth hidden layer node L and bias b L ;

[0029] S13: Generate equivalent to the hidden layer parameters To satisfy the following geometry control strategy with block increments:

[0030]

[0031] in, is obtained from the hidden parameters by changing the activation function, Δ k represents the newly added hidden node block in the kth iteration, Indicated in L-Δ k The network residual at the node time, express

[0032] S14: Calculate output weights Among them, H L =[h1,h2,…,h L ], represents the generalized inverse, Y is the discharge flow output matrix; the residual is e L =H L β L -Y, when the residual meets the tolerance error ε or the number of hidden layer nodes reaches Lmax No more nodes are added, and the modeling is completed. So far, a block incremental geometric construction network model f = H L β L .

[0033] S2: Build a sample library based on historical parameter data, use expert experience and knowledge to infer and judge historical data, and collect historical fault cases; Step S2 is as follows:

[0034] Collect and organize historical process parameters, valve opening and discharge flow data; use data cleaning technology to eliminate outliers and fill in data gaps, and build a sample library based on historical parameter data; on this basis, combine the experience and knowledge of experts to conduct in-depth analysis of the data and obtain a scoring matrix.

[0035] S3: Use the cloud model optimized by cloud entropy to classify the working condition evaluation level, combine game theory and expert scoring matrix to integrate subjective weight and objective weight to obtain the final evaluation result, and judge whether the current working condition meets the expected working condition; Step S3 is as follows:

[0036] like Figure 3 As shown, the present invention uses the cloud model to convert the evaluation results of the working conditions into numerical values ​​within [0,10] in a quantitative description manner, and divides them into 5 level intervals (excellent, good, medium, poor, and extremely poor), and the corresponding numerical intervals are [8,10], [6,8), [4,6), [2,4), and [0,2), respectively, to reflect the current discharge flow state;

[0037] Treating the class boundary as a dual-constraint space [C min ,C max ], calculate the expected Ex, which is the minimum boundary value C min With the maximum boundary value C max The mean of the sum of ;

[0038]

[0039] Solve the entropy value En according to the "3En" criterion * :

[0040]

[0041] Solve the entropy value En according to the "50% correlation" criterion ** ;

[0042]

[0043] Set the super entropy He to a fixed value t, which is usually no greater than 1 / 10 of En; introduce the cloud entropy optimization method to solve the entropy value En;

[0044] Assuming that a certain evaluation index has n evaluation levels, the objective function of the cloud entropy optimization method is to minimize the difference in certainty under each evaluation level:

[0045]

[0046] Among them, μ is the degree of certainty corresponding to the production of cloud droplets based on the digital characteristics of the cloud model after cloud entropy optimization; is the minimum value of the degree of certainty corresponding to the cloud droplets generated by the digital characteristics of the cloud model based on the “3En” criterion; is the maximum value of the certainty corresponding to the cloud droplets generated by the digital features of the cloud model based on the “50% correlation degree” criterion; the calculation formula for the cloud droplet certainty μ is:

[0047] En′=NROM(En,He 2 )

[0048]

[0049] Where En′ is a Gaussian random number generated with En as the expectation and super entropy He as the standard deviation;

[0050] The hierarchical analysis method is used to grade the evaluation indicators on the basis of expert reasoning, and the weights of each level are accurately analyzed qualitatively and quantitatively through statistical mathematical methods. Suppose the scoring matrix S = (s uv ) m×r , the specific implementation steps are as follows:

[0051] Calculate the eigenvalue of S and obtain the maximum eigenvalue λ of the judgment matrix max ;

[0052] Calculate each weight index:

[0053]

[0054] Normalize the weights and let the subjective weight be ω1={ω 1,1 ,ω 1,2 ,...,ω 1,m}, then the subjective weight ω of the u-th indicator 1,u for:

[0055]

[0056] Where η u is the weight of the u-th indicator;

[0057] The AHP method determines the reliability of the assigned weights by calculating the consistency ratio. If it is greater than 0.1, the weights need to be reallocated:

[0058]

[0059] In the formula, R I is a random index, C I is the consistency index, specifically

[0060] The entropy weight method is used to measure the discrete degree of the decision value with entropy value. The weight is assigned according to different entropy values, which can effectively avoid subjective influence and objectively reflect the data characteristics of the weight. The specific implementation steps of the entropy weight method are as follows:

[0061] For the score matrix S = (s uv ) m×r Standardize according to the following formula and calculate the proportion of the vth sample value under the uth indicator to the indicator, denoted as P = (p uv ) m×r :

[0062]

[0063] Where S max The highest score, S min The lowest score is a uv is the element in the reference entropy weight matrix obtained after normalization:

[0064] Calculate the entropy value e of the u-th indicator u :

[0065]

[0066] In the formula, l = 1 / ln(n) > 0, the entropy value e of any j-th index is j ≥0;

[0067] Let the objective weight be {ω 2，1 ,ω 2,2 ,...,ω 2,m}, the objective weight of the u-th indicator can be calculated as follows:

[0068]

[0069] Game theory is used to combine weights to obtain more reasonable indicator weights. The implementation method is as follows:

[0070] Construct a weight vector set W = {ω1, ω2}, whose weight coefficient is β = {β1, β2}, where ω1 is the subjective weight, ω2 is the objective weight, β1 is the subjective weight coefficient, and β2 is the objective weight coefficient. Suppose that any linear combination of weight vectors is expressed as follows:

[0071]

[0072] Minimize ωu With ω p,u The deviation between is the objective function, that is,

[0073]

[0074] Where u represents the weight vector of the uth column, ω p,u The transpose of

[0075] The linear equations corresponding to the optimal first-order derivative condition are:

[0076] The weight coefficient β p Normalize it:

[0077]

[0078] Calculate the final combined weight of the u-th indicator:

[0079]

[0080] According to the above method, the combined weights are used to calculate the final scores G of different experts and the average is taken. It is used as a cloud droplet to calculate the certainty of the cloud model with five evaluation levels according to the cloud model fitness calculation formula; finally, the evaluation level corresponding to the maximum certainty value is taken as the final evaluation result.

[0081] S4: If the current working condition does not meet the expected working condition, a c-means cloud clustering intelligent search strategy is used to search for the data most similar to the expected working condition from the sample library, and the corresponding valve opening value is output as the optimized operation mode; step S4 is as follows:

[0082] Through the optimization operation expert experience stored in the historical sample library, the c-means cloud clustering algorithm is used to narrow the optimization sample space range, and then pattern matching is performed, and the valve opening corresponding to the data with the largest correlation coefficient is used as the final output; the specific implementation steps are as follows:

[0083] S41: Set the initial value d=0 and randomly initialize the membership matrix where u ik is the membership of sample k to the cluster center, n is the number of samples, and c is the number of clusters;

[0084] S42: According to the maximum membership principle, the sample set g = {g1, g2, ...g N}Through U b Divided into c different clusters, and calculated according to the one-dimensional inverse cloud model Where N is the number of collected data, Ex is the expected value, En is the entropy value, and He is the super entropy;

[0085] S43: Calculate clustering object function:

[0086]

[0087] in is the sample variance of the ith category of the lth cluster, and c is the number of clusters;

[0088] S44: If |J((Ex) (b+1) ,(En) (b+1) ,(He) (b+1) )-J((Ex) (b) ,(En) (b) ,(He) (b) )|<ε, then stop, ε is the given error threshold;

[0089] If the error threshold is not met, set d = d + 1 and calculate the new membership matrix of each clustering cloud model

[0090] Where En′ is a Gaussian random number generated with En as the expectation and super entropy He as the standard deviation, Ex is the expected value, and g kl (l=1,...,p) is the lth component of the kth sample, and then returns to step S42;

[0091] The similarity coefficient between the expected working condition sample and the center of each category is calculated by the following formula:

[0092]

[0093] In the formula, g i and g j are two samples, and p is the number of elements in the sample; when the similarity coefficient is closer to 1, it indicates that the consistency between the sample and the category center is higher; the category with the largest similarity coefficient with the expected working condition sample is selected, and the valve opening value corresponding to the category is used as the control output, which is used as the optimized operation mode.

[0094] S5: Optimizing control of the valve electric actuator through a multi-rate model predictive intelligent control algorithm to achieve precise control of the valve opening; step S5 is as follows:

[0095] The multi-rate discrete model of the valve opening control system is as follows:

[0096]

[0097] Among them, z is the frame cycle step number, b represents the process state; h represents the 4-20mA control input given by the host computer, o is the control valve opening output, A f , Bf , C f and D f is the system matrix, is the output disturbance;

[0098] Setting the prediction horizon of the simulated predictive control to 2 and the control horizon to 1, the state from time z+1 to time z+2 can be obtained:

[0099]

[0100] Where b f (z) is the valve opening at the current moment, and b(z+2|z) represents the estimated value of the valve opening at time z for time z+2. Due to the fast rate adoption characteristics, the system changes little between the previous and next moments, so and use Instead, the prediction model is defined as:

[0101]

[0102] In the formula, It is known that the error between the current output value and the model calculated value is:

[0103]

[0104] Then, the predicted output of the modified model is:

[0105]

[0106] In the formula, β d is the correction vector;

[0107] The performance indicator objective function is defined as:

[0108]

[0109] In the formula, is the weighting factor, diag{} represents a diagonal matrix, is the reference sequence; then the optimization problem of simulation predictive control is transformed into:

[0110]

[0111] Finally, the optimal control rate at time p is:

[0112]

[0113] In the formula, is the controller coefficient.

Claims

1. An intelligent electronically controlled valve operation optimization control method, characterized in that: The steps include: S1: Collect the actual process parameters and valve opening data in the industrial process, and use the lightweight block incremental geometry construction network algorithm to predict the discharge flow rate; S2: Build a sample library based on historical parameter data, and use expert experience and knowledge to infer and judge historical data; S3: Use the cloud model optimized by cloud entropy to classify the working condition evaluation level, combine game theory and expert scoring matrix to integrate subjective weight and objective weight to obtain the final evaluation result, and judge whether the current working condition meets the expected working condition; S4: If the current working condition does not meet the expected working condition, a c-means cloud clustering intelligent search strategy is used to search for the data most similar to the expected working condition from the sample library, and the corresponding valve opening value is output as the optimized operation mode; S5: Through the multi-rate model predictive intelligent control algorithm, the valve electric actuator is optimized to achieve precise control of the valve opening.

2. The intelligent electric control valve operation optimization control method according to claim 1 is characterized in that Step S1 is specifically as follows: A geometrically constructed network with block increments is used as the prediction model, based on the collected historical process parameters, including flow rate, pressure, temperature, and the corresponding valve opening and discharge volume data, to obtain input data for training the neural network model. And output discharge flow data Y = {y1, y2, ..., y N },in is the process parameter data collected at time N, i is the number of process parameters, y N is the corresponding discharge flow data; all data are normalized according to the following formula to obtain The specific steps for building a prediction model are as follows: S11: Initialize the maximum number of hidden layer nodes L max , the maximum number of configuration times T max , critical tolerance error ε, random parameter range [-λ max ,λ max ] and the variation step size Δλ; S12: Randomly configure T within the adaptive interval max The input weight ω of the Lth hidden layer node L and bias b L ; S13: Generate equivalent to the hidden layer parameters To satisfy the following geometry control strategy with block increments: in, is obtained from the hidden parameters by changing the activation function, Δ k represents the newly added hidden node block in the kth iteration, Indicated in L-Δ k The network residual at the node time, express S14: Calculate output weights Among them, H L =[h1,h2,…,h L ], represents the generalized inverse, Y is the discharge flow output matrix; the residual is e L =H L β L -Y, when the residual meets the tolerance error ε or the number of hidden layer nodes reaches L max No more nodes are added, and the modeling is completed. So far, a block incremental geometric construction network model f = H is obtained. L β L .

3. The intelligent electronically controlled valve operation optimization control method according to claim 1 is characterized in that Step S2 is specifically as follows: Collect and organize historical process parameters, valve opening and discharge flow data; use data cleaning technology to eliminate outliers and fill in data gaps, and build a sample library based on historical parameter data; on this basis, combine the experience and knowledge of experts to conduct in-depth analysis of the data and obtain a scoring matrix.

4. The intelligent electronically controlled valve operation optimization control method according to claim 1 is characterized in that Step S3 is as follows: Treating the class boundary as a dual-constraint space [C min ,C max ], calculate the expected Ex, which is the minimum boundary value C min With the maximum boundary value C max The mean of the sum of ; Solve the entropy value En according to the "3En" criterion * : Solve the entropy value En according to the "50% correlation" criterion ** ; Set the super entropy He to a fixed value t, which is usually no greater than 1 / 10 of En; introduce the cloud entropy optimization method to solve the entropy value En; Assuming that a certain evaluation index has n evaluation levels, the objective function of the cloud entropy optimization method is to minimize the difference in certainty under each evaluation level: Among them, μ is the degree of certainty corresponding to the production of cloud droplets based on the digital characteristics of the cloud model after cloud entropy optimization; is the minimum value of the degree of certainty corresponding to the cloud droplets generated by the digital characteristics of the cloud model based on the "3En" criterion; is the maximum value of the certainty corresponding to the cloud droplets generated by the digital features of the cloud model based on the "50% correlation" criterion; the calculation formula for the cloud droplet certainty μ is: En′=NROM(En,He 2 , Where En′ is a Gaussian random number generated with En as the expectation and super entropy He as the standard deviation; The hierarchical analysis method is used to grade the evaluation indicators on the basis of expert reasoning, and the weights of each level are accurately analyzed qualitatively and quantitatively through statistical mathematical methods. Suppose the scoring matrix S = (s uv ) m×r , the specific implementation steps are as follows: Calculate the eigenvalue of S and obtain the maximum eigenvalue λ of the judgment matrix max ; Calculate each weight index: Normalize the weights and let the subjective weight be ω1={ω 1,1 ,ω 1,2 ,...,ω 1,m }, then the subjective weight ω of the u-th indicator 1,u for: Where η u is the weight of the u-th indicator; The AHP method determines the reliability of the assigned weights by calculating the consistency ratio. If it is greater than 0.1, the weights need to be reallocated: In the formula, R I is a random index, C I is the consistency index, specifically The entropy weight method is used to measure the discrete degree of the decision value with entropy value. The weight is assigned according to different entropy values, which can effectively avoid subjective influence and objectively reflect the data characteristics of the weight. The specific implementation steps of the entropy weight method are as follows: For the score matrix S = (s uv ) m×r Standardize according to the following formula and calculate the proportion of the vth sample value under the uth indicator to the indicator, denoted as P = (p uv ) m×r : Where S max The highest score, S min The lowest score is a uv is the element in the reference entropy weight matrix obtained after normalization: Calculate the entropy value e of the u-th indicator u : In the formula, l = 1 / ln(n) > 0, the entropy value e of any j-th index is j ≥0; Let the objective weight be {ω 2，1 ,ω 2,2 ,...,ω 2,m }, the objective weight of the u-th indicator can be calculated as follows: Game theory is used to combine weights to obtain more reasonable indicator weights. The implementation method is as follows: Construct a weight vector set W = {ω1, ω2}, whose weight coefficient is β = {β1, β2}, where ω1 is the subjective weight, ω2 is the objective weight, β1 is the subjective weight coefficient, and β2 is the objective weight coefficient. Suppose that any linear combination of weight vectors is expressed as follows: Minimize ω u With ω p,u The deviation between is the objective function, that is, Where u represents the weight vector of the uth column, ω p,u The transpose of The linear equations corresponding to the optimal first-order derivative condition are: The weight coefficient β p Normalize it: Calculate the final combined weight of the u-th indicator: According to the above method, the combined weights are used to calculate the final scores G of different experts and the average is taken. It is used as a cloud droplet to calculate the certainty of the cloud model with five evaluation levels according to the cloud model fitness calculation formula; finally, the evaluation level corresponding to the maximum certainty value is taken as the final evaluation result.

5. The intelligent electronically controlled valve operation optimization control method according to claim 1 is characterized in that Step S4 is specifically as follows: Through the optimization operation expert experience stored in the historical sample library, the c-means cloud clustering algorithm is used to narrow the optimization sample space range, and then pattern matching is performed, and the valve opening corresponding to the data with the largest correlation coefficient is used as the final output; the specific implementation steps are as follows: S41: Set the initial value d=0 and randomly initialize the membership matrix where u ik is the membership of sample k to the cluster center, n is the number of samples, and c is the number of clusters; S42: According to the maximum membership principle, the sample set g = {g1, g2, ...g N }Through U b Divided into c different clusters, and calculated according to the one-dimensional inverse cloud model Where N is the number of collected data, Ex is the expected value, En is the entropy value, and He is the super entropy; S43: Calculate clustering object function: in is the sample variance of the ith category of the lth cluster, and c is the number of clusters; S44: If |J((Ex) (b+1) ,(En) (b+1) ,(He) (b+1) )-J((Ex) (b) ,(En) (b) ,(He) (b) )|<ε, then stop, ε is the given error threshold; If the error threshold is not met, set d = d + 1 and calculate the new membership matrix of each clustering cloud model Where En′ is a Gaussian random number generated with En as the expectation and super entropy He as the standard deviation, Ex is the expected value, and g kl (l=1,...,p) is the lth component of the kth sample, and then returns to step S42; The similarity coefficient between the expected working condition sample and the center of each category is calculated by the following formula: In the formula, g i and g j are two samples, and p is the number of elements in the sample; The closer the similarity coefficient is to 1, the higher the consistency between the sample and the category center is; the category with the largest similarity coefficient with the expected operating condition sample is selected, and the valve opening value corresponding to the category is used as the control output, which is used as the optimized operation mode.

6. The intelligent electronically controlled valve operation optimization control method according to claim 1 is characterized in that Step S5 is specifically as follows: The multi-rate discrete model of the valve opening control system is as follows: Among them, z is the frame cycle step number, b represents the process state; h represents the 4-20mA control input given by the host computer, o is the control valve opening output, A f , B f , C f and D f is the system matrix, is the output disturbance; Setting the prediction horizon of the simulated predictive control to 2 and the control horizon to 1, the state from time z+1 to time z+2 can be obtained: Where b f (z) is the valve opening at the current moment, b(z+2|z) represents the estimated valve opening at time z for time z+2; Due to the fast rate adoption characteristics, the system changes little between the previous and next moments. and use to replace; Then, the prediction model is defined as: In the formula, It is known that the error between the current output value and the model calculated value is: Then, the predicted output of the modified model is: In the formula, β d is the correction vector; The performance indicator objective function is defined as: In the formula, is the weighting factor, diag{} represents a diagonal matrix, is the reference sequence; then the optimization problem of simulation predictive control is transformed into: Finally, the optimal control rate at time p is: In the formula, is the controller coefficient.

Images