System online optimization method, storage medium and computer equipment
By using the Kino polyhedron wrapping technique to represent system noise data, and cyclically optimizing complex and large-scale operating systems, the optimization problem caused by numerous uncertainties is solved, and online optimization and precise control of the system are achieved.
Patent Information
- Application Number
- CN202410974676.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-19
- Publication Date
- 2024-11-19
AI Technical Summary
Complex and large-scale operational systems such as urban sewage treatment, urban solid waste incineration, and urban water supply processes are difficult to optimize due to the large number of uncertainties and the difficulty in establishing accurate dynamic models.
The system process noise data is represented by the Chino polyhedron wrapping technique. By acquiring optimized input control data, the target system is iteratively optimized until the preset conditions are met, thereby achieving online system optimization.
It enables accurate online optimization of complex and large-scale operating systems, improving the efficiency and accuracy of system operation.
Smart Images

Figure CN118981870A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of system optimization, and in particular to a system online optimization method, storage medium and computer equipment. Background Art
[0002] Complex large-scale operation systems such as urban sewage treatment processes, urban solid waste combustion processes, and urban water supply processes have too many uncertain factors, which makes it difficult to establish accurate dynamic models and design controllers. In addition, as the large-scale operation systems operate, the uncertain factors may change. Therefore, the optimization of large-scale operation systems in operation is extremely difficult. In summary, the existing technology has the technical defect that the optimization of large-scale operation systems in operation is difficult. Summary of the invention
[0003] The purpose of this application is to provide a system online optimization method, storage medium and computer device, which can overcome the shortcomings and deficiencies in the prior art.
[0004] A first aspect of an embodiment of the present application provides a system online optimization method, comprising:
[0005] S1: obtaining an input-state data set of a target system and system process noise data represented by a Chino polyhedron wrapping technique; wherein the input-state data set of the target system includes a plurality of input-state data, each of which includes input data and system state data;
[0006] S2: Obtaining optimized input control data according to the input-state data set and the system process noise data;
[0007] S3: Optimizing the target system according to the optimized input control data;
[0008] S4: updating the input-state data set before the last optimization of the target system according to the input-state data of the target system after the last optimization to obtain a new input-state data set;
[0009] S5: obtaining new optimized input control data according to the new input-state data set and the system process noise data;
[0010] S6: Optimizing the target system again according to the new optimized input control data;
[0011] S7: Repeat the steps S4-S6 to perform loop optimization until a preset loop optimization end condition is reached, and determine the target system optimized for the last time as the optimized system.
[0012] Further, the step S2: obtaining the step of optimizing input control data according to the input-state data set and the system process noise data comprises:
[0013] S21: obtaining a parameter matrix including a plurality of candidate linear model parameters according to the input-state data set and the system process noise data;
[0014] S22: Determine a target noise Chino polyhedron with the smallest face from the parameter matrix;
[0015] S23: Acquire optimized input control data according to the target noise Chino polyhedron.
[0016] Further, step S22: determining the target noise Chino polyhedron with the smallest face from the parameter matrix comprises:
[0017] The target noise Chino polyhedron with the smallest surface is obtained by the following formula:
[0018]
[0019] in, Represents the target noise Chino polyhedron with the smallest face; ω min Represents the smallest noise data; Zonotope.enclosepoint represents the smallest Qino polyhedron that includes the noise data.
[0020] Further, step S23: obtaining optimized input control data according to the target noise Chino polyhedron includes:
[0021] S231: Obtain the polyhedron volume of the target noise Chino polyhedron;
[0022] S232: construct a discrete dynamics model of the target system and obtain a model output of the discrete dynamics model;
[0023] S233: constructing a model predictive control cost function according to the system expected output reference, the model output, and the input data;
[0024] S234: predictive control function, solving the model predictive control cost function according to a preset prediction step length, and determining the first step predictive control input value obtained by the solution as the optimized input control data.
[0025] Further, the step of S231: obtaining the polyhedron volume of the target noise Chino polyhedron comprises:
[0026] The volume of the target noise Chino polyhedron is obtained by the following formula:
[0027]
[0028] Where V represents the volume of the target noisy Kino polyhedron, volume represents the algorithm formula for obtaining the volume of the noisy Kino polyhedron, Represents the target noise Chino polyhedron with the smallest face;
[0029] Furthermore, the model predictive control cost function is:
[0030]
[0031] Among them, J represents the objective function of the model predictive control cost function, (k+i|k) represents the predicted value at the k+ith moment, E represents the difference between the model output and the system expected output reference, Z represents the prediction interval, Q and R represent the weight matrix, and F represents the terminal error matrix.
[0032] Further, the step of S4: updating the input-state data set before the last optimization of the target system according to the input-state data of the target system after the last optimization to obtain a new input-state data set includes:
[0033] The first input-state data is deleted from the input-state data set before the last optimization of the target system, and the input-state data of the target system after the last optimization is added to the end of the input-state data set to obtain a new input-state data set.
[0034] Furthermore, the step of reaching a preset cycle optimization end condition includes:
[0035] The preset number of loop optimization times input by the user is obtained, and when the real-time loop optimization times reaches the preset number of loop optimization times, it is determined that the preset loop optimization end condition is met.
[0036] A second aspect of an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the system online optimization method described above are implemented.
[0037] A third aspect of an embodiment of the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the steps of the system online optimization method as described above when executing the computer program.
[0038] Now with regard to the prior art, the present application utilizes the input-state data set of the target system and the system process noise data represented by the Chino polyhedron wrapping technology to obtain optimized input control data, and can optimize the target system according to the optimized input control data, and then update the input-state data set before the last optimization of the target system according to the input-state data of the target system after the last optimization to obtain a new input-state data set, and then obtain new optimized input control data according to the new input-state data set and the system process noise data to optimize the target system again, and the above process can be repeated to achieve cyclic optimization until the preset cyclic optimization end condition is reached, and the target system optimized for the last time is determined as the optimized system, which can be used for the optimization of large-scale operating systems in operation to accurately achieve system online optimization.
[0039] In order to provide a clearer understanding of the present application, the specific implementation of the present application will be described below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] Figure 1 This is a flow chart of a system online optimization method according to an embodiment of the present application.
[0041] Figure 2 A schematic diagram of an online rolling identification and optimization control process of a system online optimization method according to an embodiment of the present application.
[0042] Figure 3 A schematic diagram of an optimal control input data updating process based on a time series of a system online optimization method according to an embodiment of the present application.
[0043] Figure 4 This is an algorithm flow chart of a system online optimization method according to an embodiment of the present application.
[0044] Figure 5 This is a comparison diagram between the minimum noise Chino polyhedron and the system noise Chino polyhedron in an example of a system online optimization method according to an embodiment of the present application.
[0045] Figure 6 This is a schematic diagram of the system state trajectory and MPC controller output of the system online optimization method in an example of the system online optimization method according to one embodiment of the present application. DETAILED DESCRIPTION
[0046] In order to make the objectives, technical solutions and advantages of the present application more clear, the embodiments of the present application will be further described in detail below with reference to the accompanying drawings.
[0047] It should be clear that the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the embodiments of the present application.
[0048] When the following description relates to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances. The singular forms of "a", "said" and "the" used in the present application and the appended claims are also intended to include the majority form, unless the context clearly indicates other meanings. The words "if" / "if" used herein can be interpreted as "at the time of" or "when" or "in response to determination".
[0049] In addition, in the description of this application, unless otherwise specified, "plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects are in an "or" relationship.
[0050] See also Figure 1 , which is a flow chart of a system online optimization method according to an embodiment of the present application, comprising:
[0051] S1: obtaining an input-state data set of a target system and system process noise data represented by a Chino polyhedron wrapping technique; wherein the input-state data set of the target system includes a plurality of input-state data, each of which includes input data and system state data;
[0052] S2: Obtaining optimized input control data according to the input-state data set and the system process noise data;
[0053] S3: Optimizing the target system according to the optimized input control data;
[0054] S4: updating the input-state data set before the last optimization of the target system according to the input-state data of the target system after the last optimization to obtain a new input-state data set;
[0055] S5: obtaining new optimized input control data according to the new input-state data set and the system process noise data;
[0056] S6: Optimizing the target system again according to the new optimized input control data;
[0057] S7: Repeat the steps S4-S6 to perform loop optimization until a preset loop optimization end condition is reached, and determine the target system optimized for the last time as the optimized system.
[0058] In a feasible embodiment, the step S2: obtaining the step of optimizing input control data according to the input-state data set and the system process noise data comprises:
[0059] S21: obtaining a parameter matrix including a plurality of candidate linear model parameters according to the input-state data set and the system process noise data;
[0060] S22: Determine a target noise Chino polyhedron with the smallest face from the parameter matrix;
[0061] S23: Acquire optimized input control data according to the target noise Chino polyhedron.
[0062] In a feasible embodiment, step S22: determining the target noise Chino polyhedron with the smallest face from the parameter matrix, comprises:
[0063] The target noise Chino polyhedron with the smallest surface is obtained by the following formula:
[0064]
[0065] in, Represents the target noise Chino polyhedron with the smallest face; ω min Represents the smallest noise data; zonotope.enclosepoint represents the smallest Qino polyhedron that includes the noise data.
[0066] In a feasible embodiment, step S23: obtaining optimized input control data according to the target noise Chino polyhedron includes:
[0067] S231: Obtain the polyhedron volume of the target noise Chino polyhedron;
[0068] S232: construct a discrete dynamics model of the target system and obtain a model output of the discrete dynamics model;
[0069] S233: constructing a model predictive control cost function according to the system expected output reference, the model output, and the input data;
[0070] S234: predictive control function, solving the model predictive control cost function according to a preset prediction step length, and determining the first step predictive control input value obtained by the solution as the optimized input control data.
[0071] In a feasible embodiment, the step S231: obtaining the polyhedron volume of the target noise Chino polyhedron includes:
[0072] The volume of the target noise Chino polyhedron is obtained by the following formula:
[0073]
[0074] Where V represents the volume of the target noisy Kino polyhedron, volume represents the algorithm formula for obtaining the volume of the noisy Kino polyhedron, Represents the target noise Chino polyhedron with the smallest face;
[0075] In a feasible embodiment, the model predictive control cost function is:
[0076]
[0077] Among them, J represents the objective function of the model predictive control cost function, (k+i|k) represents the predicted value at the k+ith moment, E represents the difference between the model output and the system expected output reference, Z represents the prediction interval, Q and R represent the weight matrix, and F represents the terminal error matrix.
[0078] In a feasible embodiment, the step of S4: updating the input-state data set before the last optimization of the target system according to the input-state data of the target system after the last optimization to obtain a new input-state data set includes:
[0079] The first input-state data is deleted from the input-state data set before the last optimization of the target system, and the input-state data of the target system after the last optimization is added to the end of the input-state data set to obtain a new input-state data set.
[0080] In a feasible embodiment, the step of reaching a preset cycle optimization end condition includes:
[0081] The preset number of loop optimization times input by the user is obtained, and when the real-time loop optimization times reaches the preset number of loop optimization times, it is determined that the preset loop optimization end condition is met.
[0082] Now with regard to the prior art, the present application utilizes the input-state data set of the target system and the system process noise data represented by the Chino polyhedron wrapping technology to obtain optimized input control data, and can optimize the target system according to the optimized input control data, and then update the input-state data set before the last optimization of the target system according to the input-state data of the target system after the last optimization to obtain a new input-state data set, and then obtain new optimized input control data according to the new input-state data set and the system process noise data to optimize the target system again, and the above process can be repeated to achieve cyclic optimization until the preset cyclic optimization end condition is reached, and the target system optimized for the last time is determined as the optimized system, which can be used for the optimization of large-scale operating systems in operation to accurately achieve system online optimization.
[0083] In order to facilitate understanding of the technical solution of the present application, the principle of the technical solution of the present application is provided below:
[0084] Definition Notes:
[0085] In this paper, we use Define n-dimensional Euclidean space, is a positive real number, is defined for natural numbers including 0, The definition of natural numbers does not include 0. is the set of all closed intervals [a, b], such as And a≤b and is the set of all m×n real matrices and real symmetric matrices.
[0086] Definition 1: (Minkowski Sum) The Minkowski sum of two sets of vectors A and B in Euclidean space is obtained by adding each vector in A to each vector in B:
[0087]
[0088] in, Expressed as a Minkowski sum.
[0089] Definition 2: (Zonotope) Given a center and generate vector Then the Chino polyhedron is defined as:
[0090]
[0091] use Simply expressed as Its linear mapping is obtained by the following calculation:
[0092]
[0093] Given two Chino polyhedra and Its Minkowski and It can be calculated as:
[0094]
[0095] and use definition Represents the difference between two Chino polyhedra.
[0096] Definition 3: (Matrixzonotope) Given a center and the generator matrix Then a matrix Qino polyhedron is defined as:
[0097]
[0098] use Represents the matrix Qino polyhedron.
[0099] Definition 4: (Intervalmatrix) An interval matrix Indicates that each matrix element is at the left limit and the right limit The interval of all possible values between:
[0100]
[0101] The conversion of a matrix Zonotope to an interval matrix can be expressed as:
[0102]
[0103] Problem description:
[0104] Consider an unknown discrete-time nonlinear dynamic system (i.e., the target system)
[0105]
[0106] in, Indicates the status of the system. represents the input of the system, Represents the unknown noise of the system. is a discrete time function. In this work, f p Assume that it is unknown and there is no information about f p This technical solution requires the construction of a dynamic system model to capture fp The dynamic behavior of the unknown system is determined and optimized. The following are the assumptions about the data.
[0107] Assumption 1: (Data) Assume that T input-state trajectories have been obtained, each containing different lengths N j , j = 1…T data set:
[0108]
[0109] The set of system states and inputs is closed, bounded, and locally connected. The assumption that the system states are bounded is reasonable with reference to physical constraints or safety criteria.
[0110] Based on Assumption 1, the data is represented by the following matrix A collection of sequences:
[0111]
[0112] U0 represents the unknown system input data matrix at time k; X0 represents the unknown system state data matrix at time k; X1 represents the unknown system state data matrix at time k+1. For simplicity, we use Represents the above data series set.
[0113] Problem 2: Using a known dataset Capture the dynamic behavior of the unknown system, construct the state space equation of the unknown system, and design a controller to optimize its control.
[0114] See also Figure 2 , in response to question 2, this application discloses an online rolling identification and optimization control process.
[0115] The online rolling identification and optimization control process includes three modules: linear model identification module, optimal control input solution module and unknown system operation module.
[0116] First, a small amount of random control input data is used to excite the unknown system and collect a set of input-state data that can effectively identify the system information. The data is transmitted to the linear model parameter identification module, in which the linear model parameter estimation represented by the set of data is realized. Parameter estimation includes two parts: (1) using the set operation between the input-state data and the system process noise data represented by the Chino polyhedron wrapping technology, the range of all possible linear model parameters of the system consistent with the set of input-state data and the noise bound is obtained; (2) from the range of the determined linear model parameters, a set of optimal linear model parameters that minimize the wrapped noise Chino polyhedron is found. Through the calculation process (1) and (2), the linear model represented by the set of data is obtained, and the local linear fitting of the nonlinear system is realized. On the basis of obtaining the local linear model, the model predictive control algorithm is designed to calculate the unknown system response state value under the optimal control input value at the next moment based on the local linear model by considering the objective function and constraints, and the input-state data set is updated, and then the next round of local linear model parameter identification and optimal control quantity calculation is carried out.
[0117] See also Figure 3 , in order to better describe the system online optimization method of this application, Figure 3 The optimal control input data updating process based on time series is given. Specifically, first, in 0~t k In a certain period of time, a small amount of input-state operation data of the unknown system is collected. Through the aforementioned identification method, the linear system state matrix parameters representing the data are identified, and the local linear system state space model is obtained. The model predictive controller is designed using the model to calculate t k+1 The optimal control input value u at the moment k+1 , the next moment detects the unknown system state value under the optimal control input When a new set of input-state data is obtained, the data set is updated, that is, the data corresponding to the first moment in the previous set of data is discarded, and the new data is added at the end to form a new round of control input-state data set, and a new round of calculation is started. This is repeated, and finally the modeling and optimization control of the system are realized.
[0118] Next, the specific identification method and MPC (model predictive control) algorithm at each moment of the system online optimization method proposed in this application will be given.
[0119] Specific identification method for each moment:
[0120] Assume that a set of state-input data is obtained through experiments And the following data identification valid information conditions must be met.
[0121] Proposition 1: If the input-state data If the following conditions are met, the data set is considered to be informative for system identification:
[0122]
[0123] Therefore, if the data satisfies the previous formula, then there exists a right inverse matrix [V1 V2] that satisfies:
[0124]
[0125] Proposition 1 shows that It needs to be full rank, and for unknown process disturbances, it can be regarded as noise, and ω is used to represent noise data. Then for T, different lengths N j , j = 1, ...T, the superimposed noise matrix W can be expressed as:
[0126] W=[ω (1) (0)…ω (1) (N1-1)…ω (T) (0)…ω (T) (N k -1)]
[0127] In order to effectively represent noise data, this application uses the Chino polyhedron wrapping generation set technology, which can well represent different types of noise data. As shown in the form of Definition 2, it is expressed as Then, the superposition noise matrix W can be represented by the matrix Qino polyhedron shown in Definition 3, which is recorded as
[0128]
[0129]
[0130] in,
[0131]
[0132] Obviously, the matrix Qino polyhedron is composed of multiple noisy Chino polyhedra Then the matrix pair [AB] containing all the possibilities of the system can be expressed as:
[0133]
[0134] Next, the noise matrix Qino polyhedron is used to calculate the system matrix [AB] pair containing all possibilities of the set of data. The following is a method for calculating the system matrix based on the noise matrix Qino polyhedron: Given an unknown system input-state trajectory D = (U0, X) and ensuring that the matrix Full rank, then the matrix Zino polyhedron containing all possible dynamic matrices is calculated as:
[0135]
[0136] Consistent with the data D = (U0, X) and the noisy Qino polyhedron, The previous formula only uses the data D = (U0, X) to provide a purely data-based calculation method for the unknown system matrix [AB], and the previous formula uses the right inverse theory, which requires For full rank.
[0137] By using the previous formula, we can calculate t i , i=k, k+1, ..., k+m contain all possible system matrices of the unknown system at the moment. In other words, through the above set operation method, the system matrix parameters of the unknown system can be determined within a certain effective range. However, since the operation result is performed under the assumption that the noise Qino polyhedron is large enough, the obtained system matrix {[AB]} is further optimized to find more accurate system matrix parameters. The application also discloses an optimal parameter solution algorithm, which is expressed as follows:
[0138]
[0139] subject to
[0140]
[0141] Where Θ: = {, A, B} is the decision variable, volume represents the target noise Chino polyhedron Volume. In the range, by continuously searching for a set of [AB], the calculation results of X1-{AX0+BU0}} are all wrapped in the smallest Chino polyhedron Therefore, it can also be understood as finding the optimal system matrix parameters {[AB]} corresponding to the smallest Chino polyhedron containing all unknown noise data by gradually reducing the encapsulation range of the unknown noise Chino polyhedron. Therefore, by optimizing with the previous formula, it can be ensured that all unknown noises of the system are encapsulated in the smallest noise Chino polyhedron, and at the same time determine the system model matrix parameters represented by this set of data. Of course, in the optimization process, the smaller the volume of the noise Chino polyhedron, the closer it is to the actual system, and the model error is also the smallest.
[0142] Based on the above content, the present application also discloses that the previous formula can be solved by Algorithml, wherein the solving process of Algorithml is as follows:
[0143]
[0144] Among them, lines 1-4 in Algorithm1 are used to find the smallest Zino polyhedron that encloses the noise data and the optimal system matrix parameters {[AB]}. Zonotope.enclosepoint means the smallest Zino polyhedron that encloses the noise data, (:, s) means all rows in the sth column, [A k B k ] means from [AB] range Take [AB] corresponding to k, size(*) indicates the length of data(*). Algorithml can be calculated by YALMIP toolbox in MATLAB / SIMULINK programming software.
[0145] The optimal control input calculation at each moment is:
[0146] Obtain the discrete dynamics model of the unknown system (target system):
[0147]
[0148] where C is the identity matrix of appropriate dimensions and D is 0.
[0149] The objective function J of the MPC cost function (model predictive control cost function) is defined by the tracking error between the system expected output reference r and the model output y and the input u:
[0150]
[0151] In the above formula, J represents the objective function of the model predictive control cost function, (k+i|k) represents the predicted value at the k+i moment, E represents the difference between the model output and the system expected output reference, Z represents the prediction interval, Q and R represent weight matrices, and F represents the terminal error matrix.
[0152] MPC can be optimized as:
[0153]
[0154] subject to
[0155]
[0156]
[0157] ξ p (k|k)=ξ p (k)
[0158] ξp (k+1)=Aξ p (k)+Bu(k)
[0159] In the above formula, (*) lb and(*) ub represents the infimum and supremum of (*), where the optimized constraints can be given by Algorithm2 as follows:
[0160] MPC optimization control input calculation steps.
[0161]
[0162] The system online optimization method of this application combines the algorithm flow chart of Algorithm 1 and Algorithm 2 as shown in the following figure: Figure 4 shown.
[0163] In order to verify the technical solution of the present application, an experimental example of the technical solution of the present application is provided below:
[0164] Assume a two-dimensional linear discrete system, the expression formula of the system is as follows:
[0165]
[0166] The initial value of the system state is obtained from the Chino polyhedron Random values are taken from , where I is the 2D identity matrix. The input of the system is from the Chino polyhedron Random values in the system, noise data From Noise Chino Polyhedron Randomly select values and obtain 5 state trajectories of unknown system lengths of 100 through experiments.
[0167] Among them, the noise signal position of the position system, only the input-state data is known. In order to obtain the system matrix containing all the possibilities of the system (that is, the parameter matrix of multiple candidate linear model parameters), it is assumed that a sufficiently large noise Chino polyhedron is Combining the input and output data and the perturbation assumptions, according to lines 1-4 of Algorithm1, the unknown system can be calculated to contain all possible system matrix [AB] pairs. The calculation results of the left and right limits of the system matrix [AB] are as follows:
[0168]
[0169]
[0170] According to lines 5-23 of Algorithml, the system matrix is calculated using the MATLAB / YALMIP optimization toolbox as follows:
[0171]
[0172] And the minimal noise polyhedron:
[0173]
[0174] [0.0014, -0.0022, 0.0040, 0; 0.0020, 0.0012, 0, 0.0030]>
[0175] The minimum noise Chino polyhedron is obtained and system noise Chino polyhedron Comparison chart Figure 5 As shown, from Figure 5 In the figure, we can see the smallest noise Chino polyhedron obtained. The noisy Qino polyhedron given by the source system It is almost consistent, so the application can accurately estimate the unknown system matrix and achieve model building for the local linear system by finding the smallest Chino polyhedron that encapsulates the noise data.
[0176] Through the above steps, the state space model of the unknown linear discrete system can be obtained, and then the MPC optimization control algorithm is used to optimize the identified model. Among them, the control target can be set to [0; 0], the number of simulation steps is set to 300, the prediction step of the MPC optimization control algorithm is set to N = 5, and the initial state of the system is set to ξ p (0) = [5 -5] T , the controller parameters are set to: R = 0.1, Its control effect is as follows Figure 6 shown.
[0177] from Figure 6 It can be seen that the system starts running from the initial value. After the online optimization method of this application, it can accurately track the control target and achieve stable operation within a limited time. That is, this application can accurately achieve the technical effect of system online optimization.
[0178] A second aspect of an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the system online optimization method described above are implemented.
[0179] A third aspect of an embodiment of the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the steps of the system online optimization method as described above when executing the computer program.
[0180] 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 in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0181] 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 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the function selected in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 function selected in a box or multiple boxes.
[0182] 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 steps for the function selected in a box or multiple boxes.
[0183] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0184] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0185] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0186] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0187] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.
Claims
1. A system online optimization method, characterized in that: include: S1: obtaining an input-state data set of a target system and system process noise data represented by a Chino polyhedron wrapping technique; wherein the input-state data set of the target system includes a plurality of input-state data, each of which includes input data and system state data; S2: Obtaining optimized input control data according to the input-state data set and the system process noise data; S3: Optimizing the target system according to the optimized input control data; S4: updating the input-state data set before the last optimization of the target system according to the input-state data of the target system after the last optimization to obtain a new input-state data set; S5: obtaining new optimized input control data according to the new input-state data set and the system process noise data; S6: Optimizing the target system again according to the new optimized input control data; S7: Repeat the steps S4-S6 to perform loop optimization until a preset loop optimization end condition is reached, and determine the target system optimized for the last time as the optimized system.
2. The system online optimization method according to claim 1, characterized in that: The step S2: obtaining optimized input control data according to the input-state data set and the system process noise data, comprises: S21: obtaining a parameter matrix including a plurality of candidate linear model parameters according to the input-state data set and the system process noise data; S22: Determine a target noise Chino polyhedron with the smallest face from the parameter matrix; S23: Acquire optimized input control data according to the target noise Chino polyhedron.
3. The system online optimization method according to claim 2, characterized in that: Step S22: determining the target noise Chino polyhedron with the smallest face from the parameter matrix, comprising: The target noise Chino polyhedron with the smallest surface is obtained by the following formula: in, Represents the target noise Chino polyhedron with the smallest face; ω min Represents the smallest noise data; zonotope.enclosepoint represents the smallest Qino polyhedron that includes the noise data.
4. The system online optimization method according to claim 2, characterized in that: Step S23: According to the target noise Chino polyhedron, the step of obtaining optimized input control data comprises: S231: Obtain the polyhedron volume of the target noise Chino polyhedron; S232: construct a discrete dynamics model of the target system and obtain a model output of the discrete dynamics model; S233: constructing a model predictive control cost function according to the system expected output reference, the model output, and the input data; S234: predictive control function, solving the model predictive control cost function according to a preset prediction step length, and determining the first step predictive control input value obtained by the solution as the optimized input control data.
5. The system online optimization method according to claim 4, characterized in that: The step S231: obtaining the polyhedron volume of the target noise Chino polyhedron comprises: The volume of the target noise Chino polyhedron is obtained by the following formula: Where V represents the volume of the target noisy Kino polyhedron, volume represents the algorithm formula for obtaining the volume of the noisy Kino polyhedron, Represents the target noisy Qino polyhedron with the smallest face.
6. The system online optimization method according to claim 4, characterized in that: The model predictive control cost function is: Among them, J represents the objective function of the model predictive control cost function, (k+i|k) represents the predicted value at the k+ith moment, E represents the difference between the model output and the system expected output reference, Z represents the prediction interval, Q and R represent the weight matrix, and F represents the terminal error matrix.
7. The system online optimization method according to claim 1, characterized in that: The step S4: updating the input-state data set before the last optimization of the target system according to the input-state data of the target system after the last optimization to obtain a new input-state data set comprises: The first input-state data is deleted from the input-state data set before the last optimization of the target system, and the input-state data of the target system after the last optimization is added to the end of the input-state data set to obtain a new input-state data set.
8. The system online optimization method according to claim 1, characterized in that: The step of reaching the preset cycle optimization end condition includes: The preset number of loop optimization times input by the user is obtained, and when the real-time loop optimization times reaches the preset number of loop optimization times, it is determined that the preset loop optimization end condition is met.
9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the system online optimization method according to any one of claims 1 to 8 are implemented.
10. A computer device, characterized in that: The system comprises a memory, a processor and a computer program stored in the memory and executable by the processor, wherein the processor implements the steps of the system online optimization method as claimed in any one of claims 1 to 8 when executing the computer program.
Citation Information
Cited By
Online data-driven event trigger control method and system for nonlinear unmanned system under polyhedral noise
CN121541553A
