A predictive control method, system, device and medium based on distributed computing

Abstract

The present invention relates to a predictive control method, system, device and medium based on distributed computing, belonging to the field of predictive control technology. The control system used by the method is a multi-input multi-output system, which includes: obtaining the model relationship between the operating variables and the controlled variables in the control system; splitting the control system into a number of uncoupled sub-control systems based on the union-find algorithm and the model relationship; constructing the predictive control optimization proposition of each sub-control system based on the decision variables and constraints generated by the control system when optimizing the decision; performing parallel distributed solution on the predictive control optimization proposition, and integrating the optimal operating variables of each sub-control system obtained by the solution to obtain the optimal operating variables of the control system. Its beneficial effect is that the present invention adopts parallel distributed solution to the predictive control optimization proposition of each sub-control system, reduces the running time of the predictive control algorithm of the control system during the predictive control process, and improves the efficiency of predictive control.

Description

Technical Field

[0001] The present invention relates to the field of predictive control technology, and in particular to a predictive control method, system, device and medium based on distributed computing. Background Art

[0002] Predictive control, also known as model predictive control (MPC), is a multivariable control strategy that has been widely used in industrial production processes. Among them, the dynamic characteristic matrix control based on the step response sequence is widely used in process control in the chemical and petroleum industries due to its simple implementation and strong robustness. In each cycle, the dynamic characteristic matrix control needs to calculate the dynamic characteristic matrix of the relevant model. However, due to the large number of operating variables and controlled variables involved in the process industry control process, the dimension of the dynamic characteristic matrix is ​​usually large, which significantly affects the calculation efficiency of the dynamic characteristic matrix control.

[0003] In the prior art, in order to improve the computational efficiency of dynamic characteristic matrix control, a control system is generally divided into multiple sub-control systems by a subsystem partitioning method, thereby reducing the scale of the control system's operating variables and controlled variables. Subsequently, the operating variables of each sub-control system are solved by a sequential method, and the solved operating variables are applied to the original control system, thereby completing the predictive control of the industrial production process. However, in the division of sub-control systems, the adjacency matrix method is adopted, which has a high computational complexity, and for large-scale control systems, this method restricts the efficiency of control system splitting. And the sequential method is used to solve the operating variables of each sub-control system, and its solution efficiency is not high. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a predictive control method, system, device and medium based on distributed computing, which solves the technical problem of long running time of predictive control algorithms in industrial predictive control processes.

[0006] (II) Technical solution

[0007] In order to achieve the above object, the main technical solutions adopted by the present invention include:

[0008] In a first aspect, an embodiment of the present invention provides a predictive control method based on distributed computing, wherein the control system used by the method is a multi-input multi-output system, and the method comprises:

[0009] Obtain the model relationship between the operating variables and the controlled variables in the control system;

[0010] According to the union-find algorithm and the model relationship, the control system is split into a plurality of uncoupled sub-control systems;

[0011] Based on the decision variables and constraints generated by the control system when optimizing decisions, the predictive control optimization proposition of each sub-control system is constructed;

[0012] The predictive control optimization problem is solved in parallel and distributed manner, and the optimal operating variables of each sub-control system obtained by the solution are integrated to obtain the optimal operating variables of the control system.

[0013] Optionally, obtaining a model relationship between an operating variable and a controlled variable in a control system includes:

[0014] Obtain the operating variables and controlled variables in the control system;

[0015] Constructing a dynamic characteristic matrix of the control system according to the manipulated variables and the controlled variables;

[0016] According to the dynamic characteristic matrix, a model relationship between the operating variables and the controlled variables in the control system is obtained;

[0017] The dynamic characteristic matrix is ​​any one of a transfer function matrix and a step response sequence matrix.

[0018] Optionally, before splitting the control system into a plurality of uncoupled sub-control systems according to the union-find algorithm and the model relationship, the method further includes:

[0019] Obtaining the number of operating variables and the number of controlled variables in the control system;

[0020] Determining whether the quantity of the operating variables and the quantity of the controlled variables exceeds a preset quantity threshold;

[0021] When the quantity of any one of the operating variables and the controlled variables exceeds a preset quantity threshold, a union-find algorithm is called to split the control system.

[0022] Optionally, according to the union-find algorithm and the model relationship, splitting the control system into a plurality of uncoupled sub-control systems includes:

[0023] The operating variables and controlled variables in the control system are set to corresponding nodes;

[0024] Traverse the model relationships between all nodes and select at least 2 nodes as root nodes;

[0025] According to the root node, combined with the model relationship and the depth information of the multi-branch tree where the node is located, the parent nodes of the remaining nodes are set to the corresponding root nodes to obtain a multi-node tree with the same number of root nodes, which is a sub-control system.

[0026] Optionally, before constructing the predictive control optimization proposition of each sub-control system based on the decision variables and constraints generated by the control system when optimizing the decision, the method further includes:

[0027] By constructing the predictive control optimization proposition of the control system, the decision variables and constraints generated by the control system when optimizing the decision are obtained;

[0028] Among them, the predictive control optimization proposition of the control system is:

[0029] ;

[0030] In the formula, u ( i ) represents the operation variable, i =0,..., M , M represents the control time domain; represents the increment of the manipulated variable, i =0,..., M -1; y ( i ) represents the controlled variable, i =0,..., P,P represents the prediction time domain; y ref ( i ) represents the reference trajectory of the controlled variable in each cycle, i =0,..., P ; Q Represents the weight matrix of the control deviation in the controlled variable; R is the weight matrix for suppressing changes in the manipulated variables; f The predictive model dynamic equations representing the control system; represents the lower limit of the manipulated variable; represents the upper limit of the manipulated variable; Represents the lower limit of the controlled variable; Represents the upper limit of the controlled variable; Represents the incremental lower limit of the manipulated variable; Represents the upper limit of the increment of the manipulated variable.

[0031] Optionally, the predictive control optimization proposition of the sub-control system is:

[0032] ;

[0033] In the formula, N j Indicates sub-control system j The variable index set of ; z Represents a sub-control system j The index of the controlled variable; x Represents a sub-control system j The index of the operation variable; Represents the operation variable x ,in i =0,..., M , M represents the control time domain; Represents the operation variable x The increment of i =0,..., M -1; Represents the controlled variable z ,in i =0,..., P,P represents the prediction time domain; Represents the controlled variable z In each cycle of the reference trajectory, i =0,..., P ; Q Represents the weight matrix of the control deviation in the controlled variable; R A weight matrix representing the suppressed changes in the manipulated variables; Represents the operation variable x The lower limit of Represents the operation variable x The upper limit of Represents the controlled variable z The lower limit of Represents the controlled variable z The upper limit of Represents the operation variable x The lower limit of the increment; Represents the operation variable x The increment limit of .

[0034] Optionally, performing parallel distributed solving on the predictive control optimization proposition, and integrating the optimal operating variables of each sub-control system obtained by solving, to obtain the optimal operating variables of the control system includes:

[0035] Perform parallel distributed solution to the predictive control optimization proposition of the sub-control system to obtain the optimal operating variables of each sub-control system;

[0036] Integrating the optimal operating variables of each sub-control system to obtain the optimal operating variables of the control system;

[0037] The optimal operating variable of the control system is applied to the control system and combined with the controlled variable to start the control system.

[0038] In a second aspect, an embodiment of the present invention provides a predictive control system based on distributed computing, comprising:

[0039] A variable data acquisition module is used to obtain the model relationship between the operating variable and the controlled variable in the control system;

[0040] A control system splitting module, used for splitting the control system into a plurality of uncoupled sub-control systems according to a union-find algorithm and the model relationship;

[0041] A predictive control optimization proposition building module is used to build a predictive control optimization proposition for each sub-control system based on the decision variables and constraints generated by the control system when making optimization decisions;

[0042] The predictive control optimization proposition solving module is used to perform parallel distributed solving of the predictive control optimization proposition, and integrate the optimal operating variables of each sub-control system obtained by solving, so as to obtain the optimal operating variables of the control system.

[0043] In a third aspect, an embodiment of the present invention provides an electronic device, including:

[0044] processor;

[0045] A memory stores the steps of the above-mentioned predictive control method based on distributed computing for the processor to control.

[0046] In a fourth aspect, an embodiment of the present invention provides a computer-readable medium having computer-executable instructions stored thereon, and when the executable instructions are executed by a processor, the above-mentioned steps of a predictive control method based on distributed computing are implemented.

[0047] (III) Beneficial effects

[0048] The beneficial effects of the present invention are:

[0049] The present invention proposes a predictive control method based on distributed computing. Compared with the existing technology, it greatly reduces the running time of the predictive control process because it adopts parallel distributed solution to the predictive control optimization proposition of each sub-control system and then obtains the optimal operating variables of each sub-control system.

[0050] At the same time, the present invention adopts a union-find algorithm to split the control system, and the complexity of the union-find algorithm is Nlog(N), which improves the splitting efficiency of the control system while reducing the dimension of the dynamic characteristic matrix and further reduces the running time of the predictive control process. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A schematic diagram of a flow chart of a predictive control method based on distributed computing provided by one embodiment of the present invention;

[0052] Figure 2 A schematic diagram of an algorithm flow of a predictive control method based on distributed computing provided by one embodiment of the present invention;

[0053] Figure 3 A schematic diagram of an algorithm flow of a union-find algorithm provided in one embodiment of the present invention;

[0054] Figure 4 A schematic diagram of a first multitree forest state when a control system is split according to an embodiment of the present invention;

[0055] Figure 5 A schematic diagram of a second multitree forest state when a control system is split according to an embodiment of the present invention;

[0056] Figure 6 A schematic diagram of a third multitree forest state when a control system is split according to an embodiment of the present invention;

[0057] Figure 7 A schematic diagram of a fourth multitree forest state when a control system is split according to an embodiment of the present invention;

[0058] Figure 8 A schematic diagram of a fifth multitree forest state when a control system is split according to an embodiment of the present invention;

[0059] Fig. 9 A schematic diagram of a sixth multitree forest state when a control system is split according to an embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0061] Prior to this, in order to facilitate understanding of the technical solution provided by the present invention, some basic information related to the technical solution of the present application is first introduced below, and the solution of the present application can be subsequently adjusted and controlled based on the basic information.

[0062] Multiple-input multiple-output system: refers to a system with multiple inputs or outputs. Systems with a control system size greater than or equal to two inputs and two outputs are called multiple-input multiple-output systems.

[0063] Model relationship: At the controller level, if there is a model relationship (such as a step response sequence or transfer function) between a manipulated variable and the controlled variable, the change of the manipulated variable will affect the controlled variable. In this case, the controlled variable corresponding to the manipulated variable is related, that is, there is a model relationship.

[0064] Union-find: Union-find is a tree-type data structure used to handle the merging and querying of some disjoint sets.

[0065] Distributed computing: Distributed computing uses multiple processors to solve calculations, which can effectively reduce computing load and improve computing efficiency.

[0066] refer to Figure 1-Figure 3 As shown, an embodiment of the present invention proposes a predictive control method based on distributed computing, and the control system applied by the method is a multi-input and multi-output system, and the method includes: first, obtaining a model relationship between operating variables and controlled variables in the control system; second, according to a union-find algorithm and a model relationship, splitting the control system into a number of uncoupled sub-control systems; then, based on the decision variables and constraints generated by the control system when making an optimization decision, constructing a predictive control optimization proposition for each sub-control system; finally, performing parallel distributed solution on the predictive control optimization proposition, and integrating the optimal operating variables of each sub-control system obtained by solving the solution to obtain the optimal operating variables of the control system.

[0067] This embodiment adopts a technical solution of parallel distributed solution of the predictive control optimization proposition of each sub-control system to obtain the optimal operating variables of each sub-control system. Compared with the existing technology, it greatly reduces the running time of the predictive control process.

[0068] At the same time, this embodiment uses a union-find algorithm to split the control system, and the complexity of the union-find algorithm is Nlog(N), which not only reduces the dimension of the dynamic characteristic matrix, but also improves the splitting efficiency of the control system and further reduces the running time of the predictive control process.

[0069] In order to better understand the above technical solution, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a clearer and more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0070] Specifically, refer to Figure 1 and Figure 2 As shown, the embodiment of the present invention proposes a predictive control method based on distributed computing, which includes:

[0071] S1. Obtain the model relationship between the operating variables and the controlled variables in the control system.

[0072] In this embodiment, step S1 may include the following sub-steps S11-S13:

[0073] S11. Obtain the operating variables and controlled variables in the control system.

[0074] In the field of predictive control of industrial processes, based on industrial processes and control requirements, the operating variables (input variables) and controlled variables (output variables) of the control system can be set to any number. The operating variables include physical quantities such as temperature, pressure, flow, liquid level, and digital signals such as switch quantity and pulse quantity. These operating variables are converted into electrical signals or digital signals through sensors, transmitters and other devices, and then the processed signal data is transmitted to the control system for processing. The controlled variable is mainly responsible for controlling the actuators (such as motors, valves, pumps, etc.) in the industrial process according to the instructions of the control system. After receiving the controlled variables of the control system, the actuator performs corresponding actions and adjusts the operating status of the industrial process.

[0075] S12. Construct a dynamic characteristic matrix of the control system according to the manipulated variables and the controlled variables. The dynamic characteristic matrix is ​​any one of a transfer function matrix and a step response sequence matrix, and both the transfer function matrix and the step response sequence matrix can be used to calculate the model relationship between the manipulated variables and the controlled variables in the control system.

[0076] S13. According to the dynamic characteristic matrix, a model relationship between the operating variables and the controlled variables in the control system is obtained.

[0077] In a specific embodiment, a control system of a process industry uses a 4-input 3-output control system. The model relationship between the operating variables and the controlled variables in the control system is obtained through the input-output transfer function matrix of the system. The model relationship is represented by the following 4×3 matrix Mat:

[0078] (1)

[0079] In formula (1), MV1, MV2, MV3, and MV4 represent the four manipulated variables (input variables) in the control system with 4 inputs and 3 outputs, and CV1, CV2, and CV3 represent the three controlled variables (output variables) in the control system with 4 inputs and 3 outputs. The element 1 in the matrix means that the manipulated variable and the controlled variable corresponding to the row and column of the element have a model relationship.

[0080] S2. Based on the union-find algorithm and model relationship, the control system is split into several uncoupled sub-control systems.

[0081] In this embodiment, the following steps F1-F3 may be included before step S2:

[0082] F1. Obtain the number of operating variables and the number of controlled variables in the control system.

[0083] F2. Determine whether the number of the operating variables and the number of the controlled variables exceeds a preset quantity threshold.

[0084] F3. When the quantity of any one of the manipulated variables and the controlled variables exceeds a preset quantity threshold, a union-find algorithm is called to split the control system.

[0085] In a specific embodiment, the quantity threshold is set to 3. When the control system with 4 inputs and 3 outputs as described above is split into two sub-control systems, the adjacency matrix method and the union-find algorithm can be used to split the control system into two sub-control systems because the control system is small in scale and the coupling relationship of the variables is relatively clear. The matrix Mat of the sub-control system obtained by splitting using the adjacency matrix method is expressed as follows:

[0086] The matrix Mat of the first sub-control system is:

[0087] (2)

[0088] The matrix Mat of the second sub-control system is:

[0089] (3)

[0090] In this embodiment, step S2 may include the following sub-steps S21-S23:

[0091] S21. Set the operating variables and controlled variables in the control system to corresponding nodes.

[0092] S22. Traverse the model relationships between all nodes and select at least two nodes as root nodes.

[0093] S23. According to the root node, combined with the model relationship and the depth information of the multi-branch tree where the node is located, the parent nodes of the remaining nodes are set to the corresponding root nodes to obtain a multi-node tree with the same number of root nodes, which is a sub-control system.

[0094] In this embodiment, as the scale of the control system increases, the number of operating variables and controlled variables increases, the coupling relationship of the system variables is difficult to simply discover. Therefore, the present invention proposes to use a union-find algorithm to split the control system into multiple sub-control systems. The complexity of the union-find algorithm is Nlog(N). Compared with other splitting algorithms, it has high operating efficiency and short calculation time.

[0095] In a specific embodiment, according to Figure 3 The algorithm flow diagram of a union-find algorithm is shown in FIG. 1 . This embodiment uses the union-find algorithm to split the 4-input 3-output control system described above. According to the model relationship, the coupled variables are converted into multi-node trees (multi-branch trees). The root node of the multi-node tree is a sub-control system. Since the root node of the multi-node tree is always unique, it can be used to distinguish each sub-control system. The splitting process of the 4-input 3-output control system is as follows:

[0096] (1) Set the manipulated variables and controlled variables in the 4-input 3-output control system to corresponding nodes, such as Figure 4 shown.

[0097] (2) Traverse the model relationships between all nodes and select the manipulated variables MV1 and MV3 in the control system as root nodes.

[0098] (3) According to the model relationship between the manipulated variable MV1 and the controlled variable CV1, the parent node of the node of the controlled variable CV1 is set to the node of the manipulated variable MV1, that is, the manipulated variable MV1 is the root node of the controlled variable CV1, such as Figure 5 shown.

[0099] (4) According to the model relationship between the manipulated variable MV1 and the controlled variable CV2, the parent node of the node of the controlled variable CV2 is set to the node of the manipulated variable MV1, that is, the manipulated variable MV1 is the root node of the controlled variable CV2, as shown in Figure 6 shown.

[0100] (5) Since the depth of the multitree where the manipulated variable MV1 is located is greater than the depth of the multitree where the manipulated variable MV2 is located, and the manipulated variable MV1 is the root node of the controlled variable CV1, after combining the model relationship between the manipulated variable MV2 and the controlled variable CV1, the parent node of the node of the manipulated variable MV2 is set to the node of the manipulated variable MV1, as follows: Figure 7 This can reduce the number of layers of the multi-branch tree, thereby reducing the time to query the root node.

[0101] (6) According to the model relationship between the manipulated variable MV3 and the controlled variable CV3, the parent node of the node of the controlled variable CV3 is set to the node of the manipulated variable MV3, that is, the manipulated variable MV3 is the root node of the controlled variable CV3, as shown in Figure 8 shown.

[0102] (7) Since the depth of the multitree where the manipulated variable MV3 is located is greater than the depth of the multitree where the manipulated variable MV4 is located, and the manipulated variable MV3 is the root node of the controlled variable CV3, after combining the model relationship between the manipulated variable MV4 and the controlled variable CV3, the parent node of the node of the manipulated variable MV4 is set to the node of the manipulated variable MV3, as shown in the following example: Fig. 9 shown.

[0103] S3. Based on the decision variables and constraints generated by the control system when optimizing decisions, construct the predictive control optimization proposition for each sub-control system.

[0104] In this embodiment, before step S3, the following steps are also included:

[0105] By constructing the predictive control optimization proposition of the control system, the decision variables and constraints generated by the control system when making the optimization decision are obtained.

[0106] Among them, the predictive control optimization proposition of the control system is:

[0107] (4)

[0108] (4) In the formula, u ( i ) represents the operation variable, i =0,..., M , M represents the control time domain; represents the increment of the manipulated variable, i =0,..., M -1; y ( i ) represents the controlled variable, i =0,..., P,P represents the prediction time domain; y ref ( i ) represents the reference trajectory of the controlled variable in each cycle, i =0,..., P ; Q Represents the weight matrix of the control deviation in the controlled variable; R is the weight matrix for suppressing changes in the manipulated variables; f The predictive model dynamic equations representing the control system; represents the lower limit of the manipulated variable; represents the upper limit of the manipulated variable; Represents the lower limit of the controlled variable; Represents the upper limit of the controlled variable; Represents the incremental lower limit of the manipulated variable; Represents the upper limit of the increment of the manipulated variable.

[0109] In each cycle of the control system, the decision variables of the predictive control optimization proposition of the control system are L*M The constraints are K * P + L * M +2*L *( M +1)+ K *( P +1) L represents the number of manipulated variables in the control system, K Represents the number of controlled variables in the control system.

[0110] In this embodiment, the predictive control optimization proposition of the sub-control system is:

[0111] (5)

[0112] (5) In the formula, N j Indicates sub-control system j The variable index set of ; z Represents a sub-control system j The index of the controlled variable; x Represents a sub-control system j The index of the operation variable; Represents the operation variable x ,in i =0,..., M , M represents the control time domain; Represents the operation variable x The increment of i =0,..., M -1; Represents the controlled variable z ,in i =0,..., P,P represents the prediction time domain; Represents the controlled variable z In each cycle of the reference trajectory, i =0,..., P ; Q Represents the weight matrix of the control deviation in the controlled variable; R A weight matrix representing the suppressed changes in the manipulated variables; Represents the operation variable x The lower limit of Represents the operation variable x The upper limit of Represents the controlled variable z The lower limit of Represents the controlled variable z The upper limit of Represents the operation variable x The lower limit of the increment; Represents the operation variable x The increment limit of .

[0113] In each cycle of the sub-control system, the variable index set in the predictive control optimization proposition of the sub-control system exists K j The manipulated variables and L j controlled variables, the constraints of the predictive control optimization proposition of the sub-control system are K j * P + L j * M +2* L j *( M +1)+ K j *( P +1)

[0114] In a specific embodiment, a predictive control optimization proposition is constructed for two sub-control systems split from the above-mentioned 4-input 3-output control system.

[0115] The predictive control optimization proposition of the first sub-control system is:

[0116] (6)

[0117] (6) In the formula, represents the operation variable with sequence number 1, i =0,..., M , M represents the control time domain; represents the increment of the operation variable with sequence number 1, i =0,..., M -1; represents the operation variable with sequence number 2, i =0,..., M ; represents the increment of the operation variable with sequence number 2, i =0,..., M -1; represents the controlled variable with serial number 1, i =0,..., P , P represents the prediction time domain; Represents the reference trajectory of the controlled variable with serial number 1 in each cycle, i =0,..., P ; represents the controlled variable with serial number 2, i =0,..., P ; Represents the reference trajectory of the controlled variable numbered 2 in each cycle,i =0,..., P ; Represents the weight coefficient of the control deviation in the controlled variable with serial number 1; Represents the weight coefficient for suppressing changes in the manipulated variable with sequence number 1; Represents the weight coefficient of the control deviation in the controlled variable with serial number 2; Represents the weight coefficient for suppressing changes in the manipulated variable with sequence number 2; Represents the lower limit of the operating variable with sequence number 1; represents the upper limit of the manipulated variable; Represents the lower limit of the controlled variable with serial number 1; Represents the upper limit of the controlled variable with serial number 1; The increment lower limit of the operating variable with sequence number 1; Represents the increment upper limit of the operation variable with sequence number 1; Represents the lower limit of the operating variable with sequence number 2; Represents the upper limit of the operating variable with sequence number 2; Represents the lower limit of the controlled variable with serial number 2; Represents the upper limit of the controlled variable with serial number 2; Represents the lower limit of the increment of the operating variable with sequence number 2; Represents the increment upper limit of the operation variable with sequence number 2.

[0118] The predictive control optimization proposition of the second sub-control system is:

[0119] (7)

[0120] (7) In the formula, represents the operation variable with sequence number 3, i =0,..., M ; represents the increment of the operand number 3, i =0,..., M -1; represents the operation variable with sequence number 4, i =0,..., M ; represents the increment of the operation variable numbered 4, i =0,..., M -1; represents the controlled variable with serial number 3, i =0,..., P ; Represents the reference trajectory of the controlled variable numbered 3 in each cycle, i =0,..., P; Represents the weight coefficient of the control deviation in the controlled variable with serial number 3; Represents the weight coefficient for suppressing changes in the manipulated variable with sequence number 3; Represents the weight coefficient for suppressing changes in the manipulated variable with sequence number 3; Represents the lower limit of the controlled variable with serial number 3; Represents the upper limit of the controlled variable with serial number 3; Represents the lower limit of the operating variable with sequence number 3; Represents the upper limit of the operating variable with sequence number 3; Represents the lower limit of the increment of the operating variable with sequence number 3; Represents the increment upper limit of the operating variable with sequence number 3; Represents the lower limit of the operating variable with sequence number 4; Represents the upper limit of the operating variable with sequence number 4; Represents the lower limit of the increment of the operating variable with sequence number 4; Represents the increment upper limit of the manipulated variable number 4.

[0121] S4. Perform parallel distributed solution to the predictive control optimization problem, and integrate the optimal operating variables of each sub-control system to obtain the optimal operating variables of the control system.

[0122] In this embodiment, step S4 may include the following sub-steps S41-S43:

[0123] S41. Perform parallel distributed solutions to the predictive control optimization problems of the sub-control systems to obtain the optimal operating variables of each sub-control system.

[0124] S42. Integrate the optimal operating variables of each sub-control system to obtain the optimal operating variables of the control system.

[0125] S43, applying the optimal operating variables of the control system to the control system, and combining them with the controlled variables to start the control system.

[0126] Compared with the original control system, the sub-control system involves fewer variables, so the dimension of the dynamic characteristic matrix is ​​lower and the control prediction time is shorter. In addition, when solving the optimization problem, a parallel non-iterative calculation method is used so that in each cycle of the control system, all sub-control systems solve their own predictive control optimization problems at the same time. For example, assuming that the solution time of the sub-control system is T n , the total solution time of the predictive control optimization problem of the control system is T = max ( T 1,..., Tn ). It can be further seen that the total solution time depends on the calculation time of the largest sub-control system, which further reduces the running time of the predictive control process, and the parallel distributed solution only requires all sub-controllers to solve the optimization problem once in each cycle, without the need for iteration, thereby greatly improving the computational efficiency.

[0127] In addition, an embodiment of the present invention further proposes a predictive control system based on distributed computing, which includes:

[0128] A variable data acquisition module is used to obtain the model relationship between the operating variable and the controlled variable in the control system;

[0129] The control system splitting module is used to split the control system into several uncoupled sub-control systems based on the union-find algorithm and model relationship;

[0130] A predictive control optimization proposition building module is used to build a predictive control optimization proposition for each sub-control system based on the decision variables and constraints generated by the control system when making optimization decisions;

[0131] The predictive control optimization proposition solving module is used to perform parallel distributed solutions to the predictive control optimization proposition, and integrate the optimal operating variables of each sub-control system obtained by the solution to obtain the optimal operating variables of the control system.

[0132] In addition, an embodiment of the present invention further provides an electronic device, which includes: a processor; and a memory storing steps for the processor to control the above-mentioned predictive control method based on distributed computing.

[0133] It is important to understand that the processor, as the core component of the electronic device, undertakes key tasks such as data processing and instruction execution. Its powerful computing power and efficient processing speed provide a strong guarantee for the overall performance of the device. The memory plays an important role in data storage. In electronic devices, the memory not only has large-capacity and high-speed storage characteristics, but more importantly, it is specifically used to store a predictive control method based on distributed computing. These method steps are precisely controlled by the processor to achieve accurate prediction of the control strategy of the control system. This design not only improves the data processing efficiency of the device, but also enhances its adaptability and stability in complex application scenarios.

[0134] Therefore, the electronic device proposed in this embodiment, through the close cooperation between the processor and the memory, realizes efficient control and precise execution of the steps of the predictive control method based on distributed computing, further enhancing the overall performance and stability of the device.

[0135] Finally, an embodiment of the present invention also proposes a computer-readable medium on which computer executable instructions are stored. When the executable instructions are executed by the processor, the above-mentioned predictive control method steps based on distributed computing are implemented. This computer-readable medium not only provides a convenient and efficient solution for predictive control of the control system, but also ensures the accuracy and reliability of the predictive control process through the executable instructions stored therein. In short, the combination of this computer-readable medium and the processor brings unprecedented convenience and accuracy to predictive control, and further promotes the development and application of related technologies.

[0136] Specifically, in the process of predicting the control strategy of a control system, the accuracy and timeliness of the prediction are crucial. The computer-readable medium stores executable instructions for implementing a predictive control method based on distributed computing, and when these instructions are executed by a processor, it can ensure that the control strategy of the control system can be accurately predicted.

[0137] In summary, the embodiments of the present invention propose a predictive control method, system, device and medium based on distributed computing. The method is first based on the model relationship between the variables in a specific control process. Then, according to the corresponding model relationship, the controlled variables are classified using a union-find algorithm, thereby splitting the multi-input and multi-output control system into multiple uncoupled small-scale sub-control systems, thereby reducing the dimension of the dynamic characteristic matrix. Finally, the predictive control optimization proposition of each sub-control system is solved in parallel and distributed manner, thereby reducing the running time of the predictive control algorithm in the predictive control process of the control system, thereby improving the efficiency of predictive control.

[0138] Since the system / device described in the above embodiments of the present invention is a system / device used to implement the method of the above embodiments of the present invention, a person skilled in the art can understand the specific structure and deformation of the system / device based on the method described in the above embodiments of the present invention, and thus will not be described in detail here. All systems / devices used in the method of the above embodiments of the present invention belong to the scope of protection of the present invention.

[0139] It should be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention may take 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 code.

[0140] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions.

[0141] It should be noted that in the description of the present invention, the word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The present invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. The use of the words first, second, third, etc., is only for convenience of expression and does not indicate any order. These words may be understood as part of the name of the component.

[0142] In addition, it should be noted that, in the description of this specification, the description of the terms "one embodiment", "some embodiments", "embodiment", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0143] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments after obtaining the basic inventive concepts.

[0144] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the invention.

Claims

1. A predictive control method based on distributed computing, characterized in that: The control system used in the method is a multi-input multi-output system, and the method comprises: Obtain the model relationship between the operating variables and the controlled variables in the control system; According to the union-find algorithm and the model relationship, the control system is split into a plurality of uncoupled sub-control systems; Based on the decision variables and constraints generated by the control system when optimizing decisions, the predictive control optimization proposition of each sub-control system is constructed; Performing parallel distributed solutions to the predictive control optimization propositions of all sub-control systems, and integrating the optimal operating variables of all sub-control systems obtained by the solutions into one operating variable, which is the optimal operating variable of the control system; Among them, according to the union-find algorithm and the model relationship, the control system is split into several uncoupled sub-control systems, including: setting the operating variables and controlled variables in the control system to corresponding nodes; traversing the model relationship between all nodes, and selecting at least 2 nodes as root nodes; according to the root node, combined with the model relationship and the depth information of the multi-branch tree where the node is located, the parent nodes of the remaining nodes are set to the corresponding root nodes, and a multi-node tree with the same number of root nodes is obtained, and the multi-node tree is a sub-control system.

2. The method according to claim 1, characterized in that Obtaining the model relationship between the operating variables and the controlled variables in the control system includes: Obtain the operating variables and controlled variables in the control system; Constructing a dynamic characteristic matrix of the control system according to the manipulated variables and the controlled variables; According to the dynamic characteristic matrix, a model relationship between the operating variables and the controlled variables in the control system is obtained; The dynamic characteristic matrix is ​​any one of a transfer function matrix and a step response sequence matrix.

3. The method according to claim 1, characterized in that Before splitting the control system into a plurality of uncoupled sub-control systems according to the union-find algorithm and the model relationship, the method further includes: Obtaining the number of operating variables and the number of controlled variables in the control system; Determining whether the quantity of the operating variables and the quantity of the controlled variables exceeds a preset quantity threshold; When the quantity of any one of the operating variables and the controlled variables exceeds a preset quantity threshold, a union-find algorithm is called to split the control system.

4. The method according to claim 1, characterized in that Before constructing the predictive control optimization proposition of each sub-control system based on the decision variables and constraints generated by the control system when optimizing the decision, it also includes: By constructing the predictive control optimization proposition of the control system, the decision variables and constraints generated by the control system when optimizing the decision are obtained; Among them, the predictive control optimization proposition of the control system is: ； In the formula, u ( i ) represents the operation variable, i =0,..., M , M represents the control time domain; represents the increment of the manipulated variable, i =0,..., M -1; y ( i ) represents the controlled variable, i =0,..., P,P represents the prediction time domain; y ref ( i ) represents the reference trajectory of the controlled variable in each cycle, i =0,..., P ; Q Represents the weight matrix of the control deviation in the controlled variable; R is the weight matrix for suppressing changes in the manipulated variables; f The predictive model dynamic equations representing the control system; represents the lower limit of the manipulated variable; represents the upper limit of the manipulated variable; Represents the lower limit of the controlled variable; Represents the upper limit of the controlled variable; Represents the incremental lower limit of the manipulated variable; Represents the upper limit of the increment of the manipulated variable.

5. The method according to claim 1, characterized in that The predictive control optimization proposition of the sub-control system is: ； In the formula, N j Indicates sub-control system j The variable index set of ; z Represents a sub-control system j The index of the controlled variable; x Represents a sub-control system j The index of the operation variable; Represents the operation variable x ,in i =0,..., M , M represents the control time domain; Represents the operation variable x The increment of i =0,..., M -1; Represents the controlled variable z ,in i =0,..., P,P represents the prediction time domain; Represents the controlled variable z In each cycle of the reference trajectory, i =0,..., P ; Q Represents the weight matrix of the control deviation in the controlled variable; R A weight matrix representing the suppressed changes in the manipulated variables; Represents the operation variable x The lower limit of Represents the operation variable x The upper limit of Represents the controlled variable z The lower limit of Represents the controlled variable z The upper limit of Represents the operation variable x The lower limit of the increment; Represents the operation variable x The incremental upper limit of f Represents the predictive model dynamic equations of the control system.

6. The method according to claim 1, characterized in that The predictive control optimization propositions of all sub-control systems are solved in parallel and distributed manner, and the optimal operating variables of all sub-control systems obtained by the solution are integrated into one operating variable, which is the optimal operating variable of the control system and includes: Perform parallel distributed solutions to the predictive control optimization problems of all sub-control systems to obtain the optimal operating variables of all sub-control systems; Integrating the optimal operating variables of all sub-control systems into one operating variable, where the operating variable is the optimal operating variable of the control system; The optimal operating variable of the control system is applied to the control system and combined with the controlled variable to start the control system.

7. A predictive control system based on distributed computing, characterized in that: include: A variable data acquisition module is used to obtain the model relationship between the operating variable and the controlled variable in the control system; A control system splitting module is used to split the control system into several uncoupled sub-control systems according to a union-find algorithm and the model relationship; wherein, according to the union-find algorithm and the model relationship, splitting the control system into several uncoupled sub-control systems includes: setting the operating variables and controlled variables in the control system to corresponding nodes; traversing the model relationship between all nodes, and selecting at least two nodes as root nodes; according to the root node, combining the model relationship and the depth information of the multi-branch tree where the node is located, setting the parent nodes of the remaining nodes to the corresponding root nodes, and obtaining a multi-node tree with the same number of root nodes, and the multi-node tree is a sub-control system; A predictive control optimization proposition building module is used to build a predictive control optimization proposition for each sub-control system based on the decision variables and constraints generated by the control system when making optimization decisions; The predictive control optimization proposition solving module is used to perform parallel distributed solving of the predictive control optimization propositions of all sub-control systems, and integrate the optimal operating variables of all sub-control systems obtained by solving into one operating variable, which is the optimal operating variable of the control system.

8. An electronic device, characterized in that: include: processor; A memory storing the steps of a predictive control method based on distributed computing as described in any one of claims 1-6 for the processor to control.

9. A computer-readable medium having computer-executable instructions stored thereon, characterized in that: When the executable instructions are executed by the processor, the steps of a predictive control method based on distributed computing as described in any one of claims 1 to 6 are implemented.

Images