Industrial system optimization control method, device and computer readable storage medium

Abstract

The present invention belongs to the technical field of chemical and biological process optimization and control, and relates to an industrial system optimization and control method, device and computer-readable storage medium; a mathematical model, an optimization target function and an optimization and control time domain of the industrial system are obtained, and the optimization and control time domain is divided into P sub-control time domains; a control trajectory center function, an angle function and a linear basis function are constructed; for each sub-control time domain, a preset initial center value and N-1 center values ​​are respectively combined with a preset angle vector and input into a linear basis function, and N first control trajectories of variables to be optimized are output, and a first target control trajectory of the variables to be optimized and a target center value of the control trajectory of the variables to be optimized in each sub-control time domain are generated, a preset initial angle vector and N-1 angle vectors are respectively combined with the target center value and input into a linear basis function, and N second control trajectories of the variables to be optimized are output, and a second target control trajectory of the variables to be optimized is generated, and the variables to be optimized are controlled.

Description

Technical Field

[0001] The present invention relates to the technical field of chemical and biological process optimization and control, and in particular to an industrial system optimization and control method, device and computer-readable storage medium. Background Art

[0002] Chemical and biological processes are industrial processes that transform raw materials into target products through a series of physical, chemical or biological changes and reactions. They are continuous and dynamic production processes. In the production process, the optimization goal of the industrial system is often achieved by regulating the reaction variables of the industrial system. Since different industrial systems involve different types of reactions and processes, the reaction variables and optimization goals that need to be regulated are also different. For example, the intermittent feeding biochemical reactor is a small industrial system used to realize the industrial process of biological fermentation. During the fermentation process, the product is not taken out. The fermentation state is regulated by intermittent addition of raw materials. The optimization goal of maximum output at the end of the fermentation is achieved by regulating the feeding rate; the intermittent reactor is another small industrial system used to realize the industrial process of chemical synthesis. The target product is generated by the starting material or reaction intermediate. The effect of the reaction is usually affected by temperature. Therefore, for the intermittent reactor, it is often necessary to regulate the temperature to achieve the optimization goal of maximizing the output of the target product at the end of the reaction; the catalyst reactor is to achieve the optimization goal of the highest reaction efficiency by regulating the catalyst concentration. As society pays more and more attention to high efficiency and energy conservation, how to regulate the relevant reaction variables in various industrial systems to achieve optimization goals such as maximum output and highest efficiency of industrial systems has become an important issue of concern to the industry.

[0003] In the prior art, an iterative dynamic programming algorithm (IDP) is used to generate the control trajectory of the variables to be optimized in the industrial system, so as to control the variables to be optimized to meet the optimization objective function of the industrial system, thereby achieving an improvement in production efficiency or output. Specifically, the IDP algorithm first discretizes the control time domain to obtain multiple sub-control time domains, and for each sub-control time domain, multiple control trajectories of the variables to be optimized are pre-set based on experience, and the multiple control trajectories of the variables to be optimized are input into the mathematical model of the industrial system to obtain the state value of each sub-control time domain under different control trajectories of the variables to be optimized. Then, the recursive property of dynamic programming is used to recurse from the last sub-control time domain forward, and each state value and multiple control trajectories of the variables to be optimized are input into the optimization objective function of the industrial system, and the objective function value of each state value under different control trajectories of the variables to be optimized is calculated, and the control trajectory of the variables to be optimized corresponding to the optimal objective function value is used as the optimal control trajectory of the state value, thereby obtaining the target control trajectory of each state value in each sub-control time domain, and finally, each state value in the first sub-control time domain and the optimal control trajectory of the state value are input into the optimization objective function, and the target control trajectory of the state value corresponding to the optimal objective function value is used as the target control trajectory over the entire control time domain. However, since industrial systems are highly nonlinear and time-varying, their dynamic characteristics will change with time and environmental conditions. The control trajectory of the variables to be optimized that is pre-set based on experience is often subjective and cannot fully cover the dynamic characteristics and changes of the industrial system. As a result, the target control trajectory finally obtained only represents the local optimal solution of the industrial system under specific conditions, and is not the optimal control trajectory in the entire control time domain. It cannot provide the best control effect for the variables to be optimized in the industrial system, resulting in the industrial system being unable to achieve optimization goals such as maximum output and highest efficiency.

[0004] In summary, in the existing industrial system optimization and control methods, the control trajectory of the variables to be optimized, which is pre-set based on experience, is subjective and cannot fully cover the dynamic characteristics and changes of the industrial system. The target control trajectory finally obtained cannot provide the best control effect for the variables to be optimized, resulting in the industrial system being unable to achieve optimization goals such as maximum output and highest efficiency. Summary of the invention

[0005] To this end, the technical problem to be solved by the present invention is to overcome the problem that the control trajectory of the variables to be optimized, which is pre-set based on experience in the prior art, is subjective and cannot fully cover the dynamic characteristics and changes of the industrial system, so that the target control trajectory finally obtained cannot provide the best control effect, resulting in the industrial system being unable to achieve optimization goals such as maximum output and highest efficiency.

[0006] In order to solve the above technical problems, the present invention provides an industrial system optimization and control method, comprising:

[0007] Acquire a mathematical model of the industrial system, an optimization objective function, variables to be optimized, and an optimization control time domain, and divide the optimization control time domain into P sub-control time domains;

[0008] Based on the control values ​​of the control trajectory of the variable to be optimized at the start time and the end time of each sub-control time domain, construct the control trajectory center function and angle function; based on the control trajectory center function and angle function, construct a linear basis function;

[0009] For each sub-control time domain, based on the preset initial center value and the preset center value range of the control trajectory of the variable to be optimized, N-1 center values ​​are generated; the preset initial center value and the N-1 center values ​​are respectively combined with the preset angle vector and input into the linear basis function, and the first control trajectory of the N variables to be optimized in the sub-control time domain is output;

[0010] Based on the mathematical model, the optimization objective function and the N first control trajectories of the variables to be optimized in each sub-control time domain, generate the first target control trajectory of the variables to be optimized and the target center value of the control trajectory of the variables to be optimized in each sub-control time domain;

[0011] For each sub-control time domain, based on the preset initial angle vector and the preset angle vector value range of the control trajectory of the variable to be optimized, N-1 angle vectors are generated; the preset initial angle vector and the N-1 angle vectors are respectively combined with the target center value of the control trajectory of the variable to be optimized in the sub-control time domain and input into the linear basis function, and the N second control trajectories of the variables to be optimized in the sub-control time domain are output;

[0012] Based on the mathematical model, the optimization objective function and the N second control trajectories of the variables to be optimized in each sub-control time domain, a second target control trajectory of the variables to be optimized is generated, so as to control the variables to be optimized.

[0013] Preferably, the control trajectory center function is expressed as:

[0014] ,

[0015] in, represents the central function of the control trajectory; represents the control value of the control trajectory of the variable to be optimized at the beginning of the pth sub-control time domain; represents the control value of the control trajectory of the variable to be optimized at the end of the pth sub-control time domain; ;

[0016] The control trajectory angle function is expressed as:

[0017] ,

[0018] in, Indicates The duration of the sub-control time domain; Represents the proportional constant between the control value of the variable to be optimized and the optimization control time domain;

[0019] The linear basis function is expressed as:

[0020] ,

[0021] in, represents the linear basis function; Indicates The end moment of the sub-control time domain; Indicates The starting moment of the sub-control time domain; Indicates the variable to be optimized.

[0022] Preferably, the central value generation formula of the pth sub-control time domain is:

[0023] ,

[0024] in, represents the central value generated by the p-th sub-control time domain; represents the preset initial center value of the pth sub-control time domain; represents the preset central value range of the pth sub-control time domain;

[0025] The angle vector generation formula of the pth sub-control time domain is:

[0026] ,

[0027] in, represents the angle vector generated by the p-th sub-control time domain; represents the preset initial angle vector of the p-th sub-control time domain; Represents the preset angle vector value range of the p-th sub-control time domain.

[0028] Preferably, based on the mathematical model, the optimization objective function and the N to-be-optimized variable control trajectories of each sub-control time domain, the step of generating the to-be-optimized variable target control trajectory comprises:

[0029] Step 1: inputting N to-be-optimized variable control trajectories of each sub-control time domain into the mathematical model respectively, outputting N state trajectories of the sub-control time domain, and obtaining N state values ​​of the sub-control time domain based on the value of each state trajectory at the start time of the control time domain;

[0030] Step 2: For each sub-control time domain, the center value and the angle vector are respectively selected within the preset center value range and the preset angle vector range, and the center value and the angle vector obtained each time are input into the linear basis function, and a third control trajectory of the variable to be optimized is output, and based on the M value combinations of the center value and the angle vector, the M third control trajectories of the variable to be optimized in the sub-control time domain are obtained;

[0031] Step 3: Combining each state value of the P-th sub-control time domain with the M third control trajectories of the variables to be optimized in the P-th sub-control time domain and inputting them into the optimization objective function, and taking the third control trajectory of the variables to be optimized corresponding to the maximum / minimum objective function value as the target control trajectory of the state value;

[0032] Step 4: Initialize p=P-1;

[0033] Step 5: Combining each state value of the p-th sub-control time domain with the third control trajectory of the M variables to be optimized in the P-th sub-control time domain and inputting them into the mathematical model, obtaining the M target state values ​​of the state value at the end time of the p-th sub-control time domain; calculating the Euclidean distance of each target state value with the N state values ​​of the p+1-th sub-control time domain respectively, combining the third control trajectory of the variable to be optimized corresponding to the target state value with the smallest Euclidean distance with the target control trajectory of the state value of the p+1-th sub-control time domain as the target control trajectory of the state value of the p-th sub-control time domain;

[0034] Step 6: Update p=p-1, and return to execute step 5 until p=1, and input each state value of the first sub-control time domain and the target control trajectory of the state value into the optimization objective function, and use the target control trajectory of the state value corresponding to the maximum / minimum objective function value as the target control trajectory of the variable to be optimized.

[0035] Preferably, after generating the first target control trajectory of the variable to be optimized and the target center value of the control trajectory of the variable to be optimized in each sub-control time domain, the method further includes:

[0036] The preset central value range of each sub-control time domain is reduced based on the first preset reduction factor to obtain a new preset central value range, and the target central value of the control trajectory of the variable to be optimized in each sub-control time domain is used as the preset initial central value of the sub-control time domain;

[0037] For each sub-control time domain, based on the preset initial center value of the control trajectory of the variable to be optimized and the new preset center value range, N-1 center values ​​are regenerated to iterate the N first control trajectories of the variables to be optimized in each sub-control time domain until the first preset number of iterations is reached, so as to obtain N new first control trajectories of the variables to be optimized in each sub-control time domain.

[0038] Preferably, the formula for reducing the preset central value range of each sub-control time domain based on the first preset reduction factor is:

[0039] ,

[0040] in, Indicates the new preset center value range after reduction; Indicates the preset central value range before reduction; represents the first preset reduction factor.

[0041] Preferably, after generating the second target control trajectory of the variable to be optimized, the method further includes:

[0042] Based on the second target control trajectory of the variable to be optimized, the target angle vector of the control trajectory of the variable to be optimized in each sub-control time domain is obtained;

[0043] The preset angle vector value range of each sub-control time domain is reduced based on the second preset reduction factor to obtain a new preset angle vector value range, and the target angle vector of the variable control trajectory to be optimized in each sub-control time domain is used as the preset initial angle vector of the sub-control time domain;

[0044] For each sub-control time domain, based on the preset initial angle vector of the control trajectory of the variable to be optimized and the new preset angle vector value range, N-1 angle vectors are regenerated to iterate the second control trajectory of the N variables to be optimized in each sub-control time domain until the second preset number of iterations is reached, so as to obtain N new second control trajectories of the variables to be optimized in each sub-control time domain.

[0045] Preferably, the formula for reducing the preset angle vector value range of each sub-control time domain based on the second preset reduction factor is:

[0046] ,

[0047] in, Indicates the new preset angle vector value range after reduction; Indicates the preset angle vector value range before reduction; represents the second preset reduction factor.

[0048] The present invention also provides an industrial system optimization and control device, comprising:

[0049] A data acquisition module, used to acquire a mathematical model of the industrial system, an optimization objective function, variables to be optimized, and an optimization control time domain, and divide the optimization control time domain into P sub-control time domains;

[0050] A linear basis function construction module is used to construct a control trajectory center function and an angle function based on the control values ​​of the control trajectory of the variable to be optimized at the start time and the end time of each sub-control time domain; and to construct a linear basis function based on the control trajectory center function and the angle function;

[0051] The first control trajectory acquisition module of the variable to be optimized is used to generate N-1 center values ​​for each sub-control time domain based on the preset initial center value and the preset center value range of the control trajectory of the variable to be optimized; the preset initial center value and the N-1 center values ​​are respectively combined with the preset angle vector and input into the linear basis function, and the N first control trajectories of the variable to be optimized in the sub-control time domain are output;

[0052] A target center value acquisition module is used to generate a first target control trajectory of the variable to be optimized and a target center value of the control trajectory of the variable to be optimized in each sub-control time domain based on the mathematical model, the optimization objective function and the N first control trajectories of the variables to be optimized in each sub-control time domain;

[0053] The module for acquiring the second control trajectory of the variable to be optimized is used to generate N-1 angle vectors for each sub-control time domain based on the preset initial angle vector and the preset angle vector value range of the control trajectory of the variable to be optimized; the preset initial angle vector and the N-1 angle vectors are respectively combined with the target center value of the control trajectory of the variable to be optimized in the sub-control time domain and input into the linear basis function, and output the N second control trajectories of the variable to be optimized in the sub-control time domain;

[0054] The variable control module is used to generate a second target control trajectory of the variable to be optimized based on the mathematical model, the optimization objective function and the N second control trajectories of the variables to be optimized in each sub-control time domain, so as to control the variables to be optimized.

[0055] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned industrial system optimization and control method are implemented.

[0056] The present application constructs a center function and an angle function based on the control values ​​of the control trajectory of the variable to be optimized at the start and end times of each sub-control time domain, and constructs a linear basis function based on the center function and the angle function. Since the center function is used to characterize the center value of the control trajectory in the sub-control time domain, and the angle function is used to characterize the approximate direction of the control trajectory in the sub-control time domain, a set of determined center values ​​and angle vectors are input into the linear basis function to determine a unique control trajectory of the variable to be optimized. When obtaining the target control trajectory of the variable to be optimized, the center value is first used as a variable and the angle vector is used as a constant. For each sub-control time domain, multiple center values ​​are generated based on a preset initial center value and a preset center value range. Each center value is respectively combined with a preset angle vector to obtain multiple first control trajectories and The first target control trajectory of the variable to be optimized is obtained, so as to obtain the target central value of each sub-control time domain; then the target central value is taken as a constant and the angle vector is taken as a variable. For each sub-control time domain, multiple angle vectors are generated based on the preset initial angle vector and the preset angle vector value range, and each angle vector is combined with the target central value to obtain multiple second control trajectories of the variable to be optimized, so as to obtain the target control trajectory of the variable to be optimized corresponding to the optimal central value and the optimal angle vector; the multiple control trajectories of the variable to be optimized obtained by different combinations of central values ​​and angle vectors can cover the dynamic characteristics of the industrial system under different states, so that the target control trajectory of the variable to be optimized can more accurately control the variable to be optimized of the industrial system, so that the industrial system can achieve the optimization goal. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein:

[0058] Figure 1 Flow chart of the industrial system optimization and control method provided in this application;

[0059] Figure 2 Schematic diagram of the linear basis function construction principle provided for this application;

[0060] Figure 3 Schematic diagram of the principle of the industrial system optimization and control method provided in this application;

[0061] Figure 4 A schematic diagram of a control trajectory obtained using the method provided by the present application and a control trajectory obtained using a traditional method; wherein, Figure 4 (a) is a schematic diagram of the regulation trajectory obtained using the traditional IDP method. Figure 4 (b) is a schematic diagram of the regulation trajectory obtained using the method provided in this application;

[0062] Figure 5 A schematic diagram of another regulation trajectory obtained using the method provided by the present application and a regulation trajectory obtained using a traditional method; wherein, Figure 5 (a) is a schematic diagram of the regulation trajectory obtained using the traditional IDP method. Figure 5 (b) is a schematic diagram of the regulation trajectory obtained using the method provided in this application;

[0063] Figure 6 This is a schematic diagram of the structure of the industrial system optimization and control device provided in this application. DETAILED DESCRIPTION

[0064] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0065] See also Figure 1 , Figure 1 The present invention provides a flow chart of the industrial system optimization and control method, which specifically includes:

[0066] S10: Obtain a mathematical model of the industrial system, an optimization objective function, variables to be optimized, and an optimization control time domain, and divide the optimization control time domain into P sub-control time domains;

[0067] S20: constructing a center function and an angle function of the control trajectory based on the control values ​​of the control trajectory of the variable to be optimized at the start time and the end time of each sub-control time domain; constructing a linear basis function based on the center function and the angle function of the control trajectory;

[0068] S30: For each sub-control time domain, based on the preset initial center value and the preset center value range of the control trajectory of the variable to be optimized, generate N-1 center values; combine the preset initial center value and the N-1 center values ​​with the preset angle vector and input them into the linear basis function, and output the first control trajectory of the N variables to be optimized in the sub-control time domain;

[0069] S40: Based on the mathematical model, the optimization objective function and the N first control trajectories of the variables to be optimized in each sub-control time domain, generate the first target control trajectory of the variables to be optimized and the target center value of the control trajectory of the variables to be optimized in each sub-control time domain;

[0070] S50: for each sub-control time domain, based on the preset initial angle vector and the preset angle vector value range of the control trajectory of the variable to be optimized, generate N-1 angle vectors; combine the preset initial angle vector and the N-1 angle vectors with the target center value of the control trajectory of the variable to be optimized in the sub-control time domain, respectively, and input them into the linear basis function, and output the second control trajectory of the N variables to be optimized in the sub-control time domain;

[0071] S60: Based on the mathematical model, the optimization objective function and the second control trajectories of the N variables to be optimized in each sub-control time domain, a second target control trajectory of the variables to be optimized is generated, so as to control the variables to be optimized.

[0072] The present application constructs a center function and an angle function based on the control values ​​of the control trajectory of the variable to be optimized at the start and end times of each sub-control time domain, and constructs a linear basis function based on the center function and the angle function. Since the center function is used to characterize the center value of the control trajectory in the sub-control time domain, and the angle function is used to characterize the approximate direction of the control trajectory in the sub-control time domain, a set of determined center values ​​and angle vectors are input into the linear basis function to determine a unique control trajectory of the variable to be optimized. When obtaining the target control trajectory of the variable to be optimized, the center value is first used as a variable and the angle vector is used as a constant. For each sub-control time domain, multiple center values ​​are generated based on a preset initial center value and a preset center value range. Each center value is respectively combined with a preset angle vector to obtain multiple first control trajectories and The first target control trajectory of the variable to be optimized is obtained, so as to obtain the target central value of each sub-control time domain; then the target central value is taken as a constant and the angle vector is taken as a variable. For each sub-control time domain, multiple angle vectors are generated based on the preset initial angle vector and the preset angle vector value range, and each angle vector is combined with the target central value to obtain multiple second control trajectories of the variable to be optimized, so as to obtain the target control trajectory of the variable to be optimized corresponding to the optimal central value and the optimal angle vector; the multiple control trajectories of the variable to be optimized obtained by different combinations of central values ​​and angle vectors can cover the dynamic characteristics of the industrial system under different states, so that the target control trajectory of the variable to be optimized can more accurately control the variable to be optimized of the industrial system, so that the industrial system can achieve the optimization goal.

[0073] Specifically, the center function is the center value of the control trajectory in the sub-control time domain, which represents the general trend of the control trajectory of the variable to be optimized; and the angle function is the angle of the control trajectory in the sub-control time domain. Therefore, Figure 2 As shown, in a sub-control time domain, once the center value and the angle vector are determined, the unique control trajectory in the sub-control time domain can be determined;

[0074] In some embodiments of the present application, the control trajectory center function is expressed as:

[0075] ,

[0076] in, represents the central function of the control trajectory; represents the control value of the control trajectory of the variable to be optimized at the beginning of the pth sub-control time domain; represents the control value of the control trajectory of the variable to be optimized at the end of the pth sub-control time domain; ;

[0077] The control trajectory angle function is expressed as:

[0078] ,

[0079] in, Indicates The duration of the sub-control time domain; Represents the proportional constant between the control value of the variable to be optimized and the optimization control time domain; , Indicates the end time of the optimization control time domain, represents the start time of the optimization control time domain, Indicates the control trajectory of the variable to be optimized The control value at the moment, Indicates the control trajectory of the variable to be optimized Control value at the moment;

[0080] The linear basis function is expressed as:

[0081] ,

[0082] in, represents the linear basis function; Indicates The end moment of the sub-control time domain; Indicates The starting moment of the sub-control time domain; Indicates the variable to be optimized;

[0083] ,

[0084] .

[0085] Optionally, in some embodiments of the present application, the formula of the control trajectory of the variable to be optimized in the entire optimization and control time domain can be expressed as:

[0086] ,

[0087] in, Represents the control trajectory of the variable to be optimized in the entire optimization and control time domain; represents the control trajectory of the variable to be optimized in the p-th sub-control time domain; represents the switching function of the pth sub-control time domain;

[0088] .

[0089] Specifically, in some embodiments of the present application, the central value generation formula of the pth sub-control time domain is:

[0090] ,

[0091] in, represents the central value generated by the p-th sub-control time domain; represents the preset initial center value of the pth sub-control time domain; represents the preset central value range of the pth sub-control time domain;

[0092] The angle vector generation formula of the pth sub-control time domain is:

[0093] ,

[0094] in, represents the angle vector generated by the p-th sub-control time domain; represents the preset initial angle vector of the p-th sub-control time domain; Represents the preset angle vector value range of the p-th sub-control time domain.

[0095] Optionally, in order to make the final obtained control trajectory of the variable to be optimized optimal, in some embodiments of the present application, corresponding reduction factors are set for the central value range and the angle vector value range, respectively, so as to continuously reduce the central value range and the angle vector value range, so as to iterate the generated control trajectory of the variable to be optimized;

[0096] Specifically, after generating the first target control trajectory of the variable to be optimized and the target center value of the control trajectory of the variable to be optimized in each sub-control time domain, the following further includes:

[0097] The preset central value range of each sub-control time domain is reduced based on the first preset reduction factor to obtain a new preset central value range, and the target central value of the control trajectory of the variable to be optimized in each sub-control time domain is used as the preset initial central value of the sub-control time domain;

[0098] For each sub-control time domain, based on the preset initial center value of the control trajectory of the variable to be optimized and the new preset center value range, N-1 center values ​​are regenerated to iterate the N first control trajectories of the variables to be optimized in each sub-control time domain until the first preset number of iterations is reached, so as to obtain N new first control trajectories of the variables to be optimized in each sub-control time domain.

[0099] Specifically, the formula for reducing the preset central value range of each sub-control time domain based on the first preset reduction factor is:

[0100] ,

[0101] in, Indicates the new preset center value range after reduction; Indicates the preset central value range before reduction; represents the first preset reduction factor.

[0102] Furthermore, after generating the second target control trajectory of the variable to be optimized, the method further includes:

[0103] Based on the second target control trajectory of the variable to be optimized, the target angle vector of the control trajectory of the variable to be optimized in each sub-control time domain is obtained;

[0104] The preset angle vector value range of each sub-control time domain is reduced based on the second preset reduction factor to obtain a new preset angle vector value range, and the target angle vector of the variable control trajectory to be optimized in each sub-control time domain is used as the preset initial angle vector of the sub-control time domain;

[0105] For each sub-control time domain, based on the preset initial angle vector of the control trajectory of the variable to be optimized and the new preset angle vector value range, N-1 angle vectors are regenerated to iterate the second control trajectory of the N variables to be optimized in each sub-control time domain until the second preset number of iterations is reached, so as to obtain N new second control trajectories of the variables to be optimized in each sub-control time domain.

[0106] Specifically, the formula for reducing the preset angle vector value range of each sub-control time domain based on the second preset reduction factor is:

[0107] ,

[0108] in, Indicates the new preset angle vector value range after reduction; Indicates the preset angle vector value range before reduction; represents the second preset reduction factor.

[0109] Specifically, the present application uses an iterative dynamic programming algorithm to generate a target control trajectory of a variable to be optimized based on a mathematical model of an industrial system, an optimization objective function, and N control trajectories of variables to be optimized. The specific steps include:

[0110] Step 1: Input the N variable control trajectories to be optimized in each sub-control time domain into the mathematical model respectively, output the N state trajectories of the sub-control time domain, and obtain the N state values ​​of the sub-control time domain based on the value of each state trajectory at the start time of the control time domain;

[0111] Step 2: For each sub-control time domain, the center value and the angle vector are respectively selected within the preset center value range and the preset angle vector range, and the center value and the angle vector obtained each time are input into the linear basis function, and a third control trajectory of the variable to be optimized is output, and based on the M value combinations of the center value and the angle vector, the M third control trajectories of the variable to be optimized in the sub-control time domain are obtained;

[0112] Step 3: Combining each state value of the P-th sub-control time domain with the M third control trajectories of the variables to be optimized in the P-th sub-control time domain and inputting them into the optimization objective function, and taking the third control trajectory of the variables to be optimized corresponding to the maximum / minimum objective function value as the target control trajectory of the state value;

[0113] Optionally, the third control trajectory of the variable to be optimized corresponding to the maximum objective function value is used as the target control trajectory, or the third control trajectory of the variable to be optimized corresponding to the minimum objective function value is used as the target control trajectory, which can be determined according to the specific optimization objective function; for example, if the optimization objective function is the minimum objective function of industrial system cost, the third control trajectory of the variable to be optimized corresponding to the minimum objective function value is used as the target control trajectory; if the optimization objective function is the maximum objective function of industrial system efficiency, the third control trajectory of the variable to be optimized corresponding to the maximum objective function value is used as the target control trajectory;

[0114] Step 4: Initialize p=P-1;

[0115] Step 5: Combining each state value of the p-th sub-control time domain with the third control trajectory of the M variables to be optimized in the P-th sub-control time domain and inputting them into the mathematical model, obtaining the M target state values ​​of the state value at the end time of the p-th sub-control time domain; calculating the Euclidean distance between each target state value and the N state values ​​in the p+1-th sub-control time domain respectively, combining the third control trajectory of the variable to be optimized corresponding to the target state value with the smallest Euclidean distance with the target control trajectory of the state value in the p+1-th sub-control time domain as the target control trajectory of the state value in the p-th sub-control time domain;

[0116] Step 6: Update p=p-1 and return to step 5 until p=1. Input each state value of the first sub-control time domain and the target control trajectory of the state value into the optimization objective function, and use the target control trajectory of the state value corresponding to the maximum / minimum objective function value as the target control trajectory of the variable to be optimized.

[0117] like Figure 3The figure shows a schematic diagram of the principle of the industrial system optimization and control method provided by the present application. By discretizing the continuous process in time and space, a dynamic optimization problem is transformed into multiple discrete sub-optimization problems. According to the idea of ​​dynamic programming, all sub-problems are solved in reverse from back to front, and the discrete area is reduced by iteration to converge to the optimal solution. Figure 3 The figure shows a discrete method with 3 discrete states, P discrete times, and 3 optional control strategies for each discrete state. The Pth sub-control domain is solved first. For each discrete state point at the tf-L moment, the strategy that makes the objective function optimal is selected from the three control strategies. Then, the P-1th sub-control domain is returned. For each discrete state point at the tf-2L moment, the state at the tf-L moment is obtained according to the optional control strategy. The discrete state point at the tf-L moment that is closest to the desired state is selected. The control strategy selected for this discrete state point in the previous step is continued to be solved to obtain the objective function at the final moment. Finally, the optimal strategy for each discrete state point at the tf-2L moment is selected. The above process is repeated until the first sub-control domain, and the optimal control trajectory of this round of iteration is finally obtained.

[0118] Specifically, the present application embodiment also provides two specific examples to verify the effectiveness of the listed industrial system optimization and control method:

[0119] 1. The above industrial system optimization and control method is used to control the reaction variables of the intermittent reactor. The intermittent reactor is a small industrial system used to realize the chemical synthesis industrial process. In this industrial system, the chemical reaction of ABC is carried out, wherein A and C are starting materials or reaction intermediates, and B is the target product. Since the effect of the entire reaction is affected by temperature, the variable to be optimized in this industrial system is temperature, and its optimization target and mathematical model are expressed as follows:

[0120] ,

[0121] ,

[0122] ,

[0123] ,

[0124] ,

[0125] in, , Respectively represent the concentrations of substances A and B (mol / L), T represents the reaction temperature (K), represents the reaction time (h); the meaning of the optimization objective function is: generate a reaction temperature control trajectory to control the temperature so that the yield of the target product B reaches the maximum at the end of the reaction;

[0126] Specifically, in the embodiment of the present application, the optimization control time domain is divided into P sub-control time domains, the length of each sub-control time domain is L, the number N of the first control trajectory of the variable to be optimized and the second control trajectory of the variable to be optimized is 5, and the number M of the third control trajectory of the variable to be optimized is 5; the preset center value value range and the preset angle vector value range of each sub-control time domain are both 50, the preset initial center value of each sub-control time domain is 348, the first preset reduction factor is 0.85, the preset initial angle vector is 0, and the second preset reduction factor is 0.85;

[0127] As shown in Table 1, when the number of sub-control time domains is different, the objective function value after the temperature of the batch reactor is controlled by the target control trajectory of the variable to be optimized generated by the industrial system optimization control method (linear fitting IDP) provided in this application, and the objective function value after the temperature is controlled by the traditional constant fitting iterative dynamic programming algorithm:

[0128] Table 1

[0129]

[0130] It can be seen from Table 1 that under the same time-domain control rule, after the reaction temperature is controlled using the target control trajectory of the variable to be optimized obtained in the present application, the yield of B is greater; when the number of sub-control time domains is 10, the result after control using the method of the present application is better than the result of the traditional method when the number of sub-control time domains is 20 and 30, which fully proves the advantage of the change trend of the segment fitting curve provided by the present application.

[0131] like Figure 4 The figure shows the target control trajectory of the variable to be optimized obtained by using the industrial system optimization control method provided by the present application and the control trajectory of the variable to be optimized obtained by using the traditional method when the number of sub-control time domains is 40. Figure 4 (a) is the control trajectory obtained by the traditional method. Figure 4 (b) in the figure is the control trajectory obtained using the method provided by the present application. It can be seen from the figure that compared with the constant fitting of the traditional method, the control trajectory designed by the linear segment in the present application is more flexible and closer to the ideal control curve. By using the control trajectory obtained in the present application to control the reaction temperature of the intermittent reactor, the yield of the target product B can be maximized and the optimization goal can be achieved.

[0132] 2. The above industrial system optimization and control method is used to control the inverse variable of the intermittent feeding biochemical reactor. The intermittent feeding biochemical reactor is another small industrial system used to realize the biological fermentation industrial process. During the fermentation process, the product is not taken out, but the raw materials are added to the reactor intermittently to control the fermentation state. The variable to be optimized in this industrial system is the feeding rate, and its optimization target and mathematical model are expressed as:

[0133] ,

[0134] ,

[0135] ,

[0136] ,

[0137] ,

[0138] ,

[0139] ,

[0140] ,

[0141] ,

[0142] ,

[0143] ,

[0144] in, represents the secretory protein concentration, Represents the total protein concentration in the environment, Represents the bacterial concentration, represents the glucose concentration, represents the total volume of the reaction solution; the meaning of the optimization objective function is: generate a feeding rate control trajectory to control the feeding rate so that the product yield at the end of the reaction reaches the maximum;

[0145] Specifically, in the embodiment of the present application, the optimization control time domain is divided into P sub-control time domains, the length of each sub-control time domain is L, the number N of the first control trajectory of the variable to be optimized and the second control trajectory of the variable to be optimized is 5, and the number M of the third control trajectory of the variable to be optimized is 5; the preset center value of each sub-control time domain has a value range of 1, the preset angle vector has a value range of 80, the preset initial center value of each sub-control time domain is 0, the first preset reduction factor is 0.8, the preset initial angle vector is 0, and the second preset reduction factor is 0.8;

[0146] As shown in Table 2, when the number of sub-control time domains is different, the objective function value after the feed rate of the intermittent feeding biochemical reactor is regulated by the target control trajectory of the variable to be optimized generated by the industrial system optimization control method (linear fitting IDP) provided in this application, and the objective function value after the feed rate is regulated by the traditional constant fitting iterative dynamic programming algorithm:

[0147] Table 2

[0148]

[0149] It can be seen from Table 2 that under the same time-domain control rules, after the feed rate is regulated using the target control trajectory of the variable to be optimized obtained in the present application, the product yield is greater, which proves the advantage of the changing trend of the line segment fitting curve provided in the present application.

[0150] like Figure 5 The figure shows the target control trajectory of the variable to be optimized obtained by the industrial system optimization control method provided by the present application and the control trajectory of the variable to be optimized obtained by the traditional method when the number of sub-control time domains is 20. Figure 5 (a) is the control trajectory obtained by the traditional method. Figure 5 (b) in the figure is the control trajectory obtained using the method provided by the present application. It can be seen from the figure that for industrial processes with a high degree of nonlinearity, the method provided by the present application can still obtain a better control trajectory based on the traditional method, and then through the regulation of the response variables of the industrial system, the industrial system can achieve the corresponding optimization goals.

[0151] Based on the industrial system optimization and control method provided in the above embodiment, the embodiment of the present application also provides an industrial system optimization and control device, such as Figure 6 As shown, the device comprises:

[0152] The data acquisition module 10 is used to obtain the mathematical model of the industrial system, the optimization objective function, the variables to be optimized and the optimization control time domain, and divide the optimization control time domain into P sub-control time domains;

[0153] The linear basis function construction module 20 is used to construct the control trajectory center function and angle function based on the control value of the control trajectory of the variable to be optimized at the start time and the control value at the end time of each sub-control time domain; and construct the linear basis function based on the control trajectory center function and the angle function;

[0154] The first control trajectory acquisition module 30 of the variable to be optimized is used to generate N-1 center values ​​for each sub-control time domain based on the preset initial center value and the preset center value range of the control trajectory of the variable to be optimized; the preset initial center value and the N-1 center values ​​are respectively combined with the preset angle vector and input into the linear basis function, and the N first control trajectories of the variable to be optimized in the sub-control time domain are output;

[0155] A target center value acquisition module 40 is used to generate a first target control trajectory of the variable to be optimized and a target center value of the control trajectory of the variable to be optimized in each sub-control time domain based on a mathematical model, an optimization objective function and N first control trajectories of the variables to be optimized in each sub-control time domain;

[0156] The module 50 for acquiring the second control trajectory of the variable to be optimized is used to generate N-1 angle vectors for each sub-control time domain based on the preset initial angle vector and the preset angle vector value range of the control trajectory of the variable to be optimized; the preset initial angle vector and the N-1 angle vectors are respectively combined with the target center value of the control trajectory of the variable to be optimized in the sub-control time domain and input into the linear basis function, and output the N second control trajectories of the variable to be optimized in the sub-control time domain;

[0157] The variable control module 60 is used to generate a second target control trajectory of the variable to be optimized based on the mathematical model, the optimization objective function and the N second control trajectories of the variable to be optimized in each sub-control time domain, so as to control the variable to be optimized.

[0158] An embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned industrial system optimization and control method are implemented.

[0159] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0160] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0161] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0162] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0163] Obviously, the above embodiments are merely examples for the purpose of clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the present invention.

Claims

1. An industrial system optimization and control method, characterized in that: include: Obtaining a mathematical model of the industrial system, an optimization objective function, variables to be optimized, and an optimization control time domain, and dividing the optimization control time domain into P sub-control time domains; Based on the control values ​​of the control trajectory of the variable to be optimized at the start time and the end time of each sub-control time domain, the center function and the angle function of the control trajectory are constructed; Constructing a linear basis function based on the control trajectory center function and angle function; For each sub-control time domain, based on the preset initial center value and the preset center value range of the control trajectory of the variable to be optimized, N-1 center values ​​are generated; the preset initial center value and the N-1 center values ​​are respectively combined with the preset angle vector and input into the linear basis function, and the first control trajectory of the N variables to be optimized in the sub-control time domain is output; Based on the mathematical model, the optimization objective function and the N first control trajectories of the variables to be optimized in each sub-control time domain, generate the first target control trajectory of the variables to be optimized and the target center value of the control trajectory of the variables to be optimized in each sub-control time domain; For each sub-control time domain, based on the preset initial angle vector and the preset angle vector value range of the control trajectory of the variable to be optimized, N-1 angle vectors are generated; the preset initial angle vector and the N-1 angle vectors are respectively combined with the target center value of the control trajectory of the variable to be optimized in the sub-control time domain and input into the linear basis function, and the N second control trajectories of the variables to be optimized in the sub-control time domain are output; Based on the mathematical model, the optimization objective function and the N second control trajectories of the variables to be optimized in each sub-control time domain, a second target control trajectory of the variables to be optimized is generated, so as to control the variables to be optimized.

2. The industrial system optimization and control method according to claim 1, characterized in that: The control trajectory center function is expressed as: ， in, represents the central function of the control trajectory; represents the control value of the control trajectory of the variable to be optimized at the beginning of the pth sub-control time domain; represents the control value of the control trajectory of the variable to be optimized at the end of the pth sub-control time domain; ; The control trajectory angle function is expressed as: ， in, Indicates The duration of the sub-control time domain; Represents the proportional constant between the control value of the variable to be optimized and the optimization control time domain; The linear basis function is expressed as: ， in, represents the linear basis function; Indicates The end moment of the sub-control time domain; Indicates The starting moment of the sub-control time domain; Indicates the variable to be optimized.

3. The industrial system optimization and control method according to claim 1, characterized in that: The formula for generating the center value of the pth sub-control time domain is: ， in, represents the central value generated by the p-th sub-control time domain; represents the preset initial center value of the pth sub-control time domain; represents the preset central value range of the pth sub-control time domain; The angle vector generation formula of the pth sub-control time domain is: ， in, represents the angle vector generated by the p-th sub-control time domain; represents the preset initial angle vector of the p-th sub-control time domain; Represents the preset angle vector value range of the p-th sub-control time domain.

4. The industrial system optimization and control method according to claim 1, characterized in that: Based on the mathematical model, the optimization objective function and the N to-be-optimized variable control trajectories in each sub-control time domain, the step of generating the to-be-optimized variable target control trajectory comprises: Step 1: inputting N to-be-optimized variable control trajectories of each sub-control time domain into the mathematical model respectively, outputting N state trajectories of the sub-control time domain, and obtaining N state values ​​of the sub-control time domain based on the value of each state trajectory at the start time of the control time domain; Step 2: For each sub-control time domain, the center value and the angle vector are respectively selected within the preset center value range and the preset angle vector range, and the center value and the angle vector obtained each time are input into the linear basis function, and a third control trajectory of the variable to be optimized is output, and based on the M value combinations of the center value and the angle vector, the M third control trajectories of the variable to be optimized in the sub-control time domain are obtained; Step 3: Combining each state value of the P-th sub-control time domain with the M third control trajectories of the variables to be optimized in the P-th sub-control time domain and inputting them into the optimization objective function, and taking the third control trajectory of the variables to be optimized corresponding to the maximum / minimum objective function value as the target control trajectory of the state value; Step 4: Initialize p=P-1; Step 5: Combining each state value of the p-th sub-control time domain with the third control trajectory of the M variables to be optimized in the P-th sub-control time domain and inputting them into the mathematical model, obtaining the M target state values ​​of the state value at the end time of the p-th sub-control time domain; calculating the Euclidean distance of each target state value with the N state values ​​of the p+1-th sub-control time domain respectively, combining the third control trajectory of the variable to be optimized corresponding to the target state value with the smallest Euclidean distance with the target control trajectory of the state value of the p+1-th sub-control time domain as the target control trajectory of the state value of the p-th sub-control time domain; Step 6: Update p=p-1, and return to execute step 5 until p=1, and input each state value of the first sub-control time domain and the target control trajectory of the state value into the optimization objective function, and use the target control trajectory of the state value corresponding to the maximum / minimum objective function value as the target control trajectory of the variable to be optimized.

5. The industrial system optimization and control method according to claim 1, characterized in that: After generating the first target control trajectory of the variable to be optimized and the target center value of the control trajectory of the variable to be optimized in each sub-control time domain, it also includes: The preset central value range of each sub-control time domain is reduced based on the first preset reduction factor to obtain a new preset central value range, and the target central value of the control trajectory of the variable to be optimized in each sub-control time domain is used as the preset initial central value of the sub-control time domain; For each sub-control time domain, based on the preset initial center value of the control trajectory of the variable to be optimized and the new preset center value range, N-1 center values ​​are regenerated to iterate the N first control trajectories of the variables to be optimized in each sub-control time domain until the first preset number of iterations is reached, so as to obtain N new first control trajectories of the variables to be optimized in each sub-control time domain.

6. The industrial system optimization and control method according to claim 5, characterized in that: The formula for reducing the preset central value range of each sub-control time domain based on the first preset reduction factor is: ， in, Indicates the new preset center value range after reduction; Indicates the preset central value range before reduction; represents the first preset reduction factor.

7. The industrial system optimization and control method according to claim 1, characterized in that: After generating the second target control trajectory of the variable to be optimized, it also includes: Based on the second target control trajectory of the variable to be optimized, the target angle vector of the control trajectory of the variable to be optimized in each sub-control time domain is obtained; The preset angle vector value range of each sub-control time domain is reduced based on the second preset reduction factor to obtain a new preset angle vector value range, and the target angle vector of the variable control trajectory to be optimized in each sub-control time domain is used as the preset initial angle vector of the sub-control time domain; For each sub-control time domain, based on the preset initial angle vector of the control trajectory of the variable to be optimized and the new preset angle vector value range, N-1 angle vectors are regenerated to iterate the second control trajectory of the N variables to be optimized in each sub-control time domain until the second preset number of iterations is reached, so as to obtain N new second control trajectories of the variables to be optimized in each sub-control time domain.

8. The industrial system optimization and control method according to claim 7, characterized in that: The formula for reducing the preset angle vector value range of each sub-control time domain based on the second preset reduction factor is: ， in, Indicates the new preset angle vector value range after reduction; Indicates the preset angle vector value range before reduction; represents the second preset reduction factor.

9. An industrial system optimization and control device, characterized in that: include: A data acquisition module is used to acquire a mathematical model of the industrial system, an optimization objective function, variables to be optimized, and an optimization control time domain, and divide the optimization control time domain into P sub-control time domains; A linear basis function construction module is used to construct a control trajectory center function and an angle function based on the control values ​​of the control trajectory of the variable to be optimized at the start time and the end time of each sub-control time domain; and to construct a linear basis function based on the control trajectory center function and the angle function; The first control trajectory acquisition module of the variable to be optimized is used to generate N-1 center values ​​for each sub-control time domain based on the preset initial center value and the preset center value range of the control trajectory of the variable to be optimized; the preset initial center value and the N-1 center values ​​are respectively combined with the preset angle vector and input into the linear basis function, and the N first control trajectories of the variable to be optimized in the sub-control time domain are output; A target center value acquisition module is used to generate a first target control trajectory of the variable to be optimized and a target center value of the control trajectory of the variable to be optimized in each sub-control time domain based on the mathematical model, the optimization objective function and the N first control trajectories of the variables to be optimized in each sub-control time domain; The module for acquiring the second control trajectory of the variable to be optimized is used to generate N-1 angle vectors for each sub-control time domain based on the preset initial angle vector and the preset angle vector value range of the control trajectory of the variable to be optimized; the preset initial angle vector and the N-1 angle vectors are respectively combined with the target center value of the control trajectory of the variable to be optimized in the sub-control time domain and input into the linear basis function, and output the N second control trajectories of the variable to be optimized in the sub-control time domain; The variable control module is used to generate a second target control trajectory of the variable to be optimized based on the mathematical model, the optimization objective function and the N second control trajectories of the variables to be optimized in each sub-control time domain, so as to control the variables to be optimized.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the industrial system optimization and control method according to any one of claims 1 to 8 are implemented.

Images