A real-time control method for complex equipment driven by the fusion of data and rule models

Abstract

The present invention discloses a real-time control method for complex equipment driven by the fusion of data and rule models. The method comprises: a real-time control scheme credibility assessment module for complex equipment, which establishes a product quality prediction model for complex equipment based on historical production data and evaluates the credibility of each control scheme; a real-time control module for complex equipment based on improved simulated annealing, which generates a control scheme based on the product quality prediction model and an intelligent optimization algorithm; a real-time control module for complex equipment based on rule models, which establishes a real-time control rule model for complex equipment based on expert experience and generates a control scheme; and a multi-real-time control scheme fusion output module, which fuses and outputs different control schemes based on their credibility. The present invention can, to a certain extent, solve the problems of non-standard, untimely, and unreliable real-time control of complex equipment and improve the intelligence level of real-time control of complex equipment.

Description

Technical Field

[0001] The present invention belongs to the field of computer science, and in particular relates to a real-time control method for complex equipment driven by the fusion of data and rule models. Background Art

[0002] For large-scale industrial production, product quality and stability are core and critical to a company's competitiveness, placing higher demands on real-time control methods for complex equipment. During the continuous operation of complex equipment, the workshop environment, equipment operating status, and process parameters are all subject to dynamic changes, all of which affect the quality of the final product. Real-time control of complex equipment requires real-time adjustment of various control parameters based on product quality and the production environment during the production process to ensure high product quality and stability.

[0003] Traditional methods for controlling complex equipment rely primarily on manual control, relying on the operator's own experience and lacking quantitative adjustment rules. With the continuous advancement of information technology, companies and scholars have conducted various research on real-time control methods for complex equipment. Data-driven methods, such as neural networks and intelligent optimization algorithms, achieve real-time control of equipment. However, these methods are prone to inaccurate predictions and unstable control in situations such as sudden environmental changes and equipment anomalies. Rule-based model-driven methods establish real-time equipment control rule models based on expert experience, but the effectiveness of control depends on the reliability of this experience. Therefore, a more reliable real-time control method for complex equipment is urgently needed. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides a real-time control method for complex equipment driven by the fusion of data and rule models. The method covers the design of a credibility assessment module for real-time control schemes of complex equipment, the design of a real-time control module for complex equipment based on improved simulated annealing, the design of a real-time control module for complex equipment based on rule models, and the design of a fusion output module for multiple real-time control schemes. It can solve the problems of non-standard, untimely and unreliable real-time control of complex equipment to a certain extent, and improve the standardization and intelligence level of real-time control of complex equipment.

[0005] The present invention solves the technical problem by adopting the following technical solutions:

[0006] A real-time control method for complex equipment driven by the fusion of data and rule models, comprising:

[0007] (1) Design a credibility assessment module for real-time control schemes of complex equipment. The specific implementation is as follows:

[0008] ① Establish a prediction model based on historical data: Based on the historical production data of complex equipment, with production environment parameters E, equipment process parameters M, key control parameters C as input, and product quality indicators Q (such as product temperature, moisture content, density, etc.) as output, a complex equipment product quality prediction model is established;

[0009] ② Real-time prediction of product quality: During the production process, current production data is collected in real time, and the product quality index Q is obtained based on the quality prediction model;

[0010] ③ Model accuracy evaluation: Calculate the average relative error E between the product quality prediction results and the actual value over the past period T to evaluate the accuracy of the model, which is divided into five levels: very high, high, average, low, and very low;

[0011] ④ Control scheme credibility assessment: According to the level of model accuracy, determine the credibility μ1 of the data-driven complex equipment real-time control scheme and the credibility μ2 of the rule model-driven complex equipment real-time control scheme.

[0012] (2) Design a real-time control module for complex equipment based on improved simulated annealing. The specific implementation is as follows:

[0013] ① Generate initial solution: Take the actual key control parameter C as the initial solution of simulated annealing ω=(c1,c2,…,c n ), where c i represents the actual value of the i-th control parameter. The solution generated by the simulated annealing process corresponds to a set of values ​​of the control parameters of the complex device, that is, a real-time control solution;

[0014] ② Perturbation produces new solutions: Based on the current solution, random perturbation is performed within a certain range to produce new solutions, ω′=ω+Δω=(c1+Δc1,c2+Δc2,…,c n +Δc n );

[0015] ③ Evaluate the new solution: Calculate the product quality prediction value corresponding to the new solution based on the quality prediction model. If the product quality corresponding to the new solution is closer to the product quality target value than the old solution, it means that the new solution is better than the old solution; otherwise, it means that the old solution is better than the new solution.

[0016] ④Accept the new solution according to certain rules: If the new solution is better than the old solution, then accept the new solution and set ω = ω′; otherwise, accept the new solution with probability P0 and set ω = ω′, and reject the new solution with probability (1-P0);

[0017] ⑤ Determine whether the iteration termination condition is met: If the product quality corresponding to the current solution is close enough to the target value, that is, best -y t |<0.1ε(where o bestis the product quality prediction value corresponding to the optimal solution, y t is the product quality target value, ε is the threshold of the allowable deviation), then the iteration termination condition is met and the complex equipment control parameter value C1 corresponding to the current solution ω is returned, otherwise go to step ⑥;

[0018] ⑥ Determine whether the maximum number of iterations has been reached: If so, return the complex device real-time control parameter value C1 corresponding to the current solution ω; otherwise, return to step ②.

[0019] (3) Design a real-time control module for complex equipment based on a rule model. The specific implementation is as follows:

[0020] ① Establish rule model: Based on expert experience, establish a real-time control rule model for complex equipment. The rule model can be divided into relevance category, restriction category, priority category, control scheme category, weight category, etc.

[0021] ② Generate equipment control scheme: According to the current production environment and product quality, m equipment control schemes C are derived from the rule model. 21 、C 22 ,…C 2m ,;

[0022] ③ Obtain the weight of each solution: According to the credibility of each solution in the rule model, obtain the weight μ′ of each solution i ;

[0023] ④Control scheme weighting: weight multiple equipment control schemes according to the weighting coefficient, Get the weighted control parameter values ​​(c1, c2, ..., c n );

[0024] ⑤ Range restriction: The range of new control parameter values ​​is restricted according to the rule model, which mainly includes two aspects: each control parameter cannot exceed a certain range, i.e. c imin <c i <c imax , and the single adjustment range cannot be too large, that is, |Δc i |<Δc imax ;

[0025] ⑥Control scheme output: Returns the real-time control parameter C2 of the complex device after range limitation.

[0026] (4) Design a fusion output module for multiple real-time control schemes

[0027] ① Obtaining the control scheme: The control parameter C1 is obtained by the complex equipment real-time control module based on improved simulated annealing, and the control scheme C2 is obtained by the complex equipment real-time control module based on the rule model;

[0028] ② Control scheme fusion: Based on the credibility, different control schemes are weighted and integrated to obtain the final control parameter C = μ1C1 + μ2C2;

[0029] ③Control scheme output: According to the final control parameter C corresponding to (c1, c2, ..., c n ) to control complex equipment.

[0030] Through the above steps, the complex equipment control method driven by data and rule models can be combined to produce the final equipment control solution to achieve real-time control of complex equipment based on the fusion of data and rule models.

[0031] Beneficial effects:

[0032] The advantages of the present invention over the existing technology are as follows: Traditional complex equipment control methods are mainly based on manual control, relying on the operator's own experience and lacking quantitative adjustment rules. Both data-driven methods and rule model-driven methods can solve this problem to a certain extent. However, data-driven methods are prone to inaccurate predictions in situations such as sudden environmental changes, equipment maintenance, and weather changes, which in turn affects the control effect; rule model-driven methods rely on the reliability of expert experience and have low accuracy. The present invention can combine the advantages of data- and rule model-driven complex equipment control methods, evaluate and integrate different control schemes, and thus improve the accuracy of real-time control of complex equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a structural block diagram of a real-time control method for complex equipment driven by the fusion of data and rule models of the present invention. DETAILED DESCRIPTION

[0034] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0035] The present invention discloses a real-time control method for complex equipment driven by the fusion of data and rule models. For large-scale industrial production, the high quality and stability of products are the core and key to the competitiveness of enterprises, which also puts higher requirements on the real-time control method of complex equipment. During the continuous operation of complex equipment, the workshop environment, equipment operating status, process parameter indicators, etc. are all in dynamic changes, which all affect the quality of the final product. The real-time control of complex equipment requires adjusting the various control parameters of the equipment in real time according to the product quality and production environment during the production process to ensure the high quality and stability of the product. The method disclosed in the present invention includes the design of a credibility evaluation module for the real-time control method of complex equipment, the design of a real-time control module for complex equipment based on improved simulated annealing, and the design of a real-time control module for complex equipment based on a rule model. It can solve the problems of non-standard, untimely and unreliable real-time control of complex equipment to a certain extent, and improve the standardization and intelligence level of real-time control of complex equipment.

[0036] like Figure 1 As shown, the present invention is a real-time control method for complex equipment driven by the fusion of data and rule models. The specific implementation method is as follows:

[0037] (1) Design a credibility assessment module 1 for real-time control schemes of complex equipment. The specific implementation is as follows:

[0038] ① Establish a prediction model based on historical data: Based on the historical production data of complex equipment, with production environment parameters E (such as ambient temperature, ambient humidity, ambient atmospheric pressure, etc.), equipment process parameters M (such as speed, equipment temperature, etc.), key control parameters C (such as valve opening, hot air speed, etc.) as input, and product quality indicators Q (such as product temperature, moisture content, density, etc.) as output, a complex equipment product quality prediction model is established;

[0039] ② Real-time prediction of product quality: During the production process, current production data is collected in real time, and the product quality index Q is obtained based on the quality prediction model;

[0040] ③ Model accuracy evaluation: Calculate the average relative error E between the product quality prediction results and the actual values ​​over the past period T to evaluate the accuracy of the model. Let the current time be c and the time window be T. The average relative error E of the calculated model is:

[0041]

[0042] Among them, y c-i is the actual value of the complex equipment product quality at time (ci), c-i is the predicted value of the complex equipment product quality at time (ci). According to the size of the average relative error of the model, the accuracy of the model prediction can be measured as very high, high, average, low, and very low:

[0043]

[0044] ④ Control scheme credibility evaluation: According to the specific level of model accuracy, the credibility μ1 of the data-driven complex equipment real-time control scheme and the credibility μ2 of the rule-based model-driven complex equipment real-time control scheme are determined, as shown in Table 1.

[0045] Table 1 Correspondence table of model accuracy and control scheme credibility

[0046] Model accuracy <![CDATA[μ1]]> <![CDATA[μ2]]> Very high 1 0 Higher 0.75 0.25 generally 0.5 0.5 Lower 0.25 0.75 Very low 0 1

[0047] (2) Design a real-time control module 2 for complex equipment based on improved simulated annealing, which is specifically implemented as follows:

[0048] ① Generate initial solution: Take the actual key control parameter C as the initial solution of simulated annealing ω=(c1,c2,…,c n ), where c i represents the actual value of the i-th control parameter. The solution generated by the simulated annealing process corresponds to a set of values ​​of the control parameters of the complex device, that is, a real-time control solution;

[0049] ② Perturbation produces new solutions: Based on the current solution, random perturbation is performed within a certain range to produce new solutions, ω′=ω+Δω=(c1+Δc1,c2+Δc2,…,c n +Δc n );

[0050] ③ Evaluate the new solution: Calculate the product quality prediction value corresponding to the new solution based on the quality prediction model. If the product quality corresponding to the new solution is closer to the product quality target value than the old solution, it means that the new solution is better than the old solution; otherwise, it means that the old solution is better than the new solution.

[0051] ④Accept the new solution according to certain rules: If the new solution is better than the old solution, then accept the new solution and set ω = ω′; otherwise, accept the new solution with probability P0 and set ω = ω′, and reject the new solution with probability (1-P0);

[0052] ⑤ Determine whether the iteration termination condition is met: If the product quality corresponding to the current solution is close enough to the target value, that is, best -y t |<0.1ε, where o best is the product quality prediction value corresponding to the optimal solution, y t is the product quality target value, ε is the threshold of the allowable deviation, then the iteration termination condition is met and the real-time control parameter value C1 of the complex equipment corresponding to the current solution ω is returned. Otherwise, go to step ⑥;

[0053] ⑥ Determine whether the maximum number of iterations has been reached: If so, return the complex device real-time control parameter value C1 corresponding to the current solution ω; otherwise, return to step ②.

[0054] Complex device control has high real-time requirements, does not require finding a global optimal solution, and requires that the adjustment of control parameters be minimized during each optimization. Therefore, compared with the traditional simulated annealing algorithm, the present invention makes the following improvements: 1) The traditional simulated annealing algorithm has two loops: the outer loop slowly decreases the temperature T, and the inner loop iteratively generates and selects solutions at a fixed temperature. The present invention retains only the inner loop to accelerate the algorithm's search for qualified solutions. 2) In the traditional simulated annealing algorithm, the acceptance of differential solutions is based on the Metropolis criterion. As the iterations proceed, the probability of accepting differential solutions decreases. The present invention accepts differential solutions with a fixed small probability, reducing the algorithm's ability to escape local optimal solutions in the early stages of the iteration and the number of unnecessary iterations. 3) The traditional simulated annealing algorithm has no requirements for the selection of the initial solution and often randomly selects a point in the solution space as the initial solution. In contrast, the present invention uses the current actual control parameter value as the initial solution value. Subsequent iterations are perturbed based on the current control parameter value, so that the optimal solution obtained is as close as possible to the current control parameter value. Based on the above improvements, the improved simulated annealing algorithm has the characteristics of fewer iterations and faster solution speed compared with the traditional simulated annealing algorithm and other classic intelligent optimization algorithms, and can better meet the real-time needs of complex equipment control.

[0055] (3) Design a real-time control module 3 for complex equipment based on rule models. The specific implementation is as follows:

[0056] ① Establish a rule model: Based on expert experience, a real-time control rule model for complex equipment is established. The rule model can be divided into related categories, restriction categories, priority categories, control categories, weight categories, etc., as shown in Table 2.

[0057] Table 2 Real-time control rule model for complex equipment

[0058]

[0059] Among them, C is the equipment control parameter, Q is the product quality index, R represents the correlation, Prim represents the priority, E represents the production environment, and μ represents the weight;

[0060] ②Generate equipment control plan: According to the current production environment and product quality, the equipment control plan is derived from the rule model, namely C 2i =f(E,Q,M); Each control scheme can be specifically expressed as the value of each control parameter;

[0061] ③ Obtain the weight of each solution: According to the credibility of each solution in the rule model, obtain the weight μ′ of each solutioni ;

[0062] ④ Control scheme weighting: weight multiple equipment control schemes according to the weighting coefficient to obtain new control parameter values, that is,

[0063]

[0064] After weighting, the control parameter values ​​(c1, c2, ..., c n )

[0065] ⑤Range restriction: The new control parameter value is restricted according to the rule model. This mainly includes two aspects: each control parameter cannot exceed a certain range, and the single adjustment range cannot be too large. The expression formula is as follows:

[0066]

[0067] ⑥Control scheme output: Returns the real-time control parameter C2 of the complex device after range limitation.

[0068] (4) Design a multi-real-time control scheme fusion output module 4, which is specifically implemented as follows:

[0069] ① Obtaining the control scheme: The complex equipment real-time control module 2 based on improved simulated annealing obtains the data-driven complex equipment real-time control parameter C1, and the complex equipment real-time control module 3 based on the rule model obtains the rule-model-driven complex equipment real-time control parameter C2;

[0070] ② Control scheme fusion: Perform weighted fusion of different control schemes based on credibility, C=μ1C1+μ2C2, and obtain the final control scheme result.

[0071] ③Control scheme output: According to the control parameter values ​​(c1, c2, ..., c n ) to control complex equipment.

[0072] In summary, the present invention discloses a real-time control method for complex equipment driven by the fusion of data and rule models, including a credibility evaluation module for the real-time control scheme of complex equipment, a real-time control module design for complex equipment based on an improved simulated annealing algorithm, a real-time control module design for complex equipment based on a rule model, and a fusion output module design for the real-time control scheme of complex equipment. It can solve the problems of non-standard, untimely and unreliable real-time control of complex equipment to a certain extent, and improve the standardization and intelligence level of real-time control of complex equipment.

[0073] The contents not described in detail in the specification of the present invention belong to the prior art known to those skilled in the art.

[0074] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A real-time control method for complex equipment driven by the fusion of data and rule models, characterized in that: These include: (1) Design a credibility assessment module for real-time control schemes of complex equipment, including: ① Construction of prediction model based on historical data: Based on the historical production data of complex equipment, with production environment parameters E, equipment process parameters M, key control parameters C as input, and product quality indicators Q as output, a complex equipment product quality prediction model is established; ② Real-time prediction of product quality: During the production process, current production data is collected in real time, and the product quality index Q is obtained based on the quality prediction model; ③ Model accuracy evaluation: Calculate the average relative error E between the product quality prediction results and the actual value over the past period T to evaluate the accuracy of the model, which is divided into five levels: very high, high, average, low, and very low; ④Control scheme credibility assessment: Determine the credibility of data-driven real-time control schemes for complex equipment based on the level of model accuracy Credibility of real-time control solutions for complex equipment driven by rule models ; (2) Design a real-time control module for complex equipment based on improved simulated annealing, including: ① Generate initial solution: Use the current actual key control parameter C as the initial solution of simulated annealing ,in represents the actual value of the i-th control parameter. The solution generated by the simulated annealing process corresponds to a set of values ​​of the control parameters of the complex device, that is, a real-time control solution; ② Perturbation produces new solutions: Based on the corresponding current solution, random perturbations are performed within a certain range to produce new solutions. ; ③ Evaluate the new solution: Calculate the product quality prediction value corresponding to the new solution based on the quality prediction model. If the product quality corresponding to the new solution is closer to the product quality target value than the old solution, it means that the new solution is better than the old solution; otherwise, it means that the old solution is better than the new solution. ④Accept the new solution according to certain rules: If the new solution is better than the old solution, then accept the new solution and let Otherwise Accept the new solution for probability, let ,by Reject the new solution for probability; ⑤ Determine whether the iteration termination condition is met: If the product quality corresponding to the current solution is close enough to the target value, that is, ,in, is the product quality prediction value corresponding to the optimal solution, is the product quality target value, If the threshold of the allowed deviation is met, the iteration termination condition is met and the current solution is returned. The corresponding complex device control parameter value C1, otherwise go to step ⑥; ⑥ Determine whether the maximum number of iterations has been reached: if so, return the current solution The corresponding complex device real-time control parameter value C1, otherwise return to step ②; (3) Design a real-time control module for complex equipment based on rule models, including: ① Establish rule model: Establish real-time control rule model for complex equipment. The rule model is divided into related categories, restriction categories, priority categories, control scheme categories, and weight categories; ② Generate equipment control scheme: According to the current production environment and product quality, m equipment control schemes C are derived from the rule model. 21 、C 22 ,… C 2m , where m is a positive integer between 3 and 10; ③ Obtain the weight of each solution: According to the credibility of each solution in the rule model, obtain the weight of each solution ; ④Control scheme weighting: weight multiple equipment control schemes according to the weighting coefficient, , get the weighted control parameter value ; ⑤Range restriction: The new control parameter value is restricted according to the rule model, including two aspects: each control parameter cannot exceed a certain range , and the single adjustment range cannot be too large ; ⑥Control solution output: Returns the real-time control parameters C2 of the complex equipment after range restriction; (4) Design a multi-real-time control scheme fusion output module, including: ① Obtaining the control scheme: The control parameter C1 is obtained by the complex equipment real-time control module based on improved simulated annealing, and the control scheme C2 is obtained by the complex equipment real-time control module based on the rule model; ② Control scheme fusion: Based on the credibility, different control schemes are weighted and integrated to obtain the final control parameters. ; ③Control scheme output: According to the final control parameters Corresponding , to control complex equipment.

2. The method for real-time control of complex equipment driven by the fusion of data and rule models according to claim 1, characterized in that: The step (3) generates a device control plan by establishing a real-time control rule model for complex devices.

Images