Multi-objective optimization method and device for operation of power distribution system, and electronic equipment

Abstract

The invention discloses a multi-objective optimization method and device for operation of a power distribution system and electronic equipment, and relates to the field of power systems, and the method comprises the steps: predicting photovoltaic output power according to meteorological data through a photovoltaic power prediction model; constructing a multi-target optimization model based on the photovoltaic output power predicted by the photovoltaic power prediction model; determining the hierarchical weight of each solving target of the multi-target optimization model through an analytic hierarchy process; converting the multi-objective optimization model into a single-objective solving function according to the hierarchical weight of each solving objective; and solving the single-target solving function according to a target particle swarm optimization algorithm to obtain a regulation and control strategy of operation of the power distribution system, and the target particle swarm optimization algorithm is a particle swarm optimization algorithm for determining an initial population by using a chaotic mapping mode. According to the method and the device, the problem that safe and economical operation is difficult to consider in a power distribution system connected with a high-permeability distributed power supply due to the lack of an effective multi-target coordination mechanism and an optimization algorithm in the prior art is solved.

Description

Technical Field

[0001] This application relates to the field of power systems, and more specifically, to a multi-objective optimization method, apparatus, and electronic equipment for the operation of power distribution systems. Background Technology

[0002] With the construction and development of new power distribution systems, large-scale, decentralized, intermittent, and random distributed power sources are being connected to the power distribution system, greatly increasing the complexity and management difficulty of the power grid.

[0003] In existing technologies, although optimization scheduling models for distribution networks with a high proportion of photovoltaic power are constructed, and methods such as model predictive control, stochastic optimization, or robust optimization are used to formulate scheduling strategies with the objectives of minimizing system operating costs, minimizing grid losses, or maximizing photovoltaic absorption, they often rely on the decision-maker's prior experience for subjective weighting when faced with multiple conflicting objectives, lacking a systematic and objective trade-off mechanism. Furthermore, existing algorithms generally suffer from slow local convergence speeds, high model complexity, and difficulty in meeting the real-time requirements of online rolling optimization, making it difficult to balance safe and economical operation in distribution systems with high penetration of distributed power sources.

[0004] There is currently no effective solution to the above problems. Summary of the Invention

[0005] This application provides a multi-objective optimization method, apparatus, and electronic device for power distribution system operation, which at least solves the technical problem in the prior art that it is difficult to balance safe and economical operation in power distribution systems with high penetration of distributed power sources due to the lack of effective multi-objective coordination mechanisms and optimization algorithms.

[0006] According to one aspect of the embodiments of this application, a multi-objective optimization method for power distribution system operation is provided, comprising: predicting photovoltaic power output based on meteorological data using a photovoltaic power prediction model, wherein the photovoltaic power prediction model is a vector machine model optimized by an objective optimization algorithm, the objective optimization algorithm being used to control the prediction accuracy of the vector machine model by adjusting the penalty parameter and the kernel parameter of the radial basis function; constructing a multi-objective optimization model based on the photovoltaic power output predicted by the photovoltaic power prediction model, wherein the solution objectives of the multi-objective optimization model include grid losses, voltage deviation, and operating costs of the power distribution system; determining the hierarchical weights of each solution objective of the multi-objective optimization model using the analytic hierarchy process (AHP); converting the multi-objective optimization model into a single-objective solution function according to the hierarchical weights of each solution objective; and solving the single-objective solution function using a target particle swarm optimization algorithm to obtain the control strategy for the operation of the power distribution system, wherein the target particle swarm optimization algorithm is a particle swarm optimization algorithm that uses chaotic mapping to determine the initial population.

[0007] Optionally, before predicting photovoltaic power output based on meteorological data using the photovoltaic power prediction model, the method further includes: establishing an initial vector machine model, wherein the kernel function of the initial vector machine model is a radial basis function, the kernel parameters of the radial basis function are used to control the model complexity, and the model parameters of the initial vector machine model also include penalty parameters used to control the model's tolerance to training data errors; automatically adjusting the penalty parameters and kernel parameters in the initial vector machine model through a target optimization algorithm to obtain a target vector machine model, wherein the target optimization algorithm is used to find the optimal combination of penalty parameters and kernel parameters in the parameter space of the initial vector machine model through an iterative process by simulating the leadership structure and social behavior of a wolf pack; in each iteration, the target optimization algorithm uses the directions indicated by the three leader wolves to guide the population update to approach the optimal solution, and the target optimization algorithm gradually reduces the search range through a linear reduction strategy until convergence; and iteratively training the target vector machine model using a training dataset to obtain the photovoltaic power prediction model.

[0008] Optionally, the photovoltaic power prediction model is obtained by iteratively training the target vector machine model using a training dataset, including: acquiring historical meteorological data of the photovoltaic power station as training samples, wherein the historical meteorological data includes cloud cover information, temperature range, humidity changes, solar radiation intensity, and photovoltaic system characteristic parameters; acquiring historical photovoltaic power output data of the photovoltaic power station as training sample labels; using the training samples and training sample labels as a training dataset, and iteratively training the target vector machine model using the training dataset to obtain the photovoltaic power prediction model.

[0009] Optionally, based on the photovoltaic output power predicted by the photovoltaic power prediction model, a multi-objective optimization model is constructed, including: constructing a grid loss objective function and a voltage deviation objective function based on the photovoltaic output power predicted by the photovoltaic power prediction model, wherein the voltage deviation objective function is used to minimize the voltage deviation of each node in the distribution network, and the grid loss objective function is used to minimize the grid loss of each branch in the distribution network; constructing an operating cost objective function, wherein the operating cost objective function is used to minimize the overall operating cost of the distribution network; and using the harmonic result of the grid loss objective function, the voltage deviation objective function, and the operating cost objective function as the multi-objective optimization model.

[0010] Optionally, the parameters of the power grid loss objective function include the resistance of each branch in the distribution network, the active and reactive power flowing through the branch, and the voltage of the branch terminal node.

[0011] Optionally, the parameters of the operating cost objective function include the economic benefits of distributed resource generation and the maintenance costs of distributed resource generation. The economic benefits of distributed resource generation are determined by the distributed resource generation revenue coefficient and the distributed resource output value. The maintenance costs of distributed resource generation are determined by the distributed resource maintenance and operation cost coefficient and the distributed resource output value. Among them, distributed resources include at least photovoltaic power plants.

[0012] Optionally, the hierarchical weights of each solution objective in the multi-objective optimization model are determined using the analytic hierarchy process (AHP), including: determining the relative importance score of each solution objective in the multi-objective optimization model by comparing each solution objective with the other solution objectives in the multi-objective optimization model pairwise; constructing a judgment matrix based on the relative importance scores of each solution objective in multiple objective optimization models, where each element in the judgment matrix represents the relative importance of a pair of solution objectives in the multi-objective optimization model, and the rows and columns of the judgment matrix correspond to different solution objectives in the multi-objective optimization model; determining the eigenvectors of the judgment matrix; and using each component of the eigenvector as a hierarchical weight of a solution objective.

[0013] Optionally, after treating each component of the feature vector as a hierarchical weight of a solution objective, the method further includes: detecting the consistency ratio among the hierarchical weights of all solution objectives in the multi-objective optimization model, wherein the consistency ratio is used to determine the degree of logical consistency between the construction of the judgment matrix and the allocation of hierarchical weights; and adjusting the judgment matrix if the consistency ratio is detected to be greater than a preset threshold, until the consistency ratio is detected to be less than or equal to the preset threshold.

[0014] Optionally, the multi-objective optimization model is converted into a single-objective solution function according to the hierarchical weights of each solution objective, including: performing weighted calculations on the functions corresponding to each solution objective in the multi-objective optimization model according to the hierarchical weights of each solution objective to obtain a single-objective solution function.

[0015] Optionally, the method further includes: in the process of converting the multi-objective optimization model into a single-objective solution function according to the hierarchical weights of each solution objective, determining the objective constraints of the single-objective solution function based on all the constraints of the multi-objective optimization model, wherein the constraints of the multi-objective optimization model include: power flow equation constraints, node voltage magnitude constraints, branch thermal constraints, and distributed resource output regulation constraints.

[0016] Optionally, the control strategy for the operation of the power distribution system is obtained by solving the single-objective solution function using the target particle swarm optimization algorithm, including: generating a particle swarm at the initial position based on a chaotic sequence; generating a series of chaotic numbers using the logistic mapping in chaos theory, wherein the logistic mapping generates a pseudo-random sequence through nonlinear iteration, which is used to break the regularity of the particle swarm positions; initializing the positions of the particle swarm using the generated chaotic numbers; after initializing the positions of the particle swarm, finding the optimal solution in the solution space of the single-objective solution function using the target particle swarm optimization algorithm, wherein each particle represents a potential solution; and determining the control strategy for the operation of the power distribution system based on the optimal solution found in the solution space of the single-objective solution function, wherein the control strategy includes: the optimal output value of distributed resources, the optimal network loss and voltage deviation for the operation of the power distribution system.

[0017] Optionally, the method further includes: during the process of finding the optimal solution in the solution space of the single-objective solution function using the target particle swarm optimization algorithm, determining the fitness value of the solution represented by each particle, wherein the fitness of the solution characterizes the degree to which the solution satisfies the single-objective optimization function, and the lower the fitness value of the solution, the closer the solution is to the optimization objective of the objective function, and the greater the probability that the solution is the optimal solution.

[0018] Optionally, the method further includes: during the process of finding the optimal solution in the solution space of the single-objective solution function using the target particle swarm optimization algorithm, at each iteration of the particle swarm, calculating the fitness value of each particle in the particle swarm according to its position in the solution space; statistically analyzing the fitness value of each particle in the particle swarm and calculating the average fitness value of all particles; calculating the variance corresponding to the fitness value of the particle swarm based on the fitness value of each particle in the particle swarm and the average fitness value of all particles, wherein the variance reflects the dispersion of the fitness value of the particle swarm; if the variance is detected to be less than the target threshold, generating new particle positions using the logistic mapping in chaos theory to replace some particles in the particle swarm; if the variance is detected to be greater than or equal to the target threshold, continuing to the next iteration of the particle swarm until the number of iterations exceeds the preset number or the target particle swarm optimization algorithm is detected to have entered a convergent state.

[0019] According to another aspect of the embodiments of this application, a multi-objective optimization device for power distribution system operation is also provided, comprising: a power prediction unit, used to predict photovoltaic power output based on meteorological data using a photovoltaic power prediction model, wherein the photovoltaic power prediction model is a vector machine model optimized based on an objective optimization algorithm, and the objective optimization algorithm is used to control the prediction accuracy of the vector machine model by adjusting the penalty parameter and the kernel parameter of the radial basis function; a model construction unit, used to construct a multi-objective optimization model based on the photovoltaic power output predicted by the photovoltaic power prediction model, wherein the solution objectives of the multi-objective optimization model include grid loss, voltage deviation and operating cost of the power distribution system; a hierarchical weight determination unit, used to determine the hierarchical weights of each solution objective of the multi-objective optimization model using the analytic hierarchy process (AHP); a model conversion unit, used to convert the multi-objective optimization model into a single-objective solution function according to the hierarchical weights of each solution objective; and a control strategy determination unit, used to solve the single-objective solution function according to a target particle swarm optimization algorithm to obtain the control strategy for the operation of the power distribution system, wherein the target particle swarm optimization algorithm is a particle swarm optimization algorithm that uses chaotic mapping to determine the initial population.

[0020] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, which stores a computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located performs the above-described multi-objective optimization method for power distribution system operation.

[0021] According to another aspect of the embodiments of this application, an electronic device is also provided, including one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by one or more processors, the one or more processors cause the one or more processors to perform the above-described multi-objective optimization method for power distribution system operation.

[0022] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program or instructions, which, when executed by a processor, implement the above-described multi-objective optimization method for power distribution system operation.

[0023] In this application, the intelligent distribution network system first predicts photovoltaic (PV) power output based on meteorological data using a PV power prediction model. This PV power prediction model is a vector machine model optimized using an objective optimization algorithm. The objective optimization algorithm controls the prediction accuracy of the vector machine model by adjusting the penalty parameter and the kernel parameter of the radial basis function. Based on the PV power output predicted by the PV power prediction model, a multi-objective optimization model is constructed. The solution objectives of the multi-objective optimization model include grid losses, voltage deviation, and operating costs of the distribution system. The hierarchical weights of each solution objective in the multi-objective optimization model are determined using the analytic hierarchy process (AHP). Based on these hierarchical weights, the multi-objective optimization model is converted into a single-objective solution function. Then, the single-objective solution function is solved using a target particle swarm optimization algorithm to obtain the control strategy for the operation of the distribution system. This target particle swarm optimization algorithm is a particle swarm optimization algorithm that uses chaotic mapping to determine the initial population.

[0024] As described above, the intelligent distribution network system of this application, employing an optimized vector machine model, better adapts to the intermittency and volatility of photovoltaic power by adjusting the penalty parameters and the kernel parameters of the radial basis function. Based on this, the intelligent distribution network system constructs a multi-objective optimization model, comprehensively considering grid losses, voltage deviations, and operating costs. The hierarchical weights of each optimization objective are determined using the analytic hierarchy process (AHP), and the multi-objective optimization model is transformed into a single-objective solution function, achieving simplification and efficient solution of complex optimization problems. The intelligent distribution network system uses a particle swarm optimization algorithm based on chaotic mapping to solve the single-objective solution function. The initial population is generated through chaotic mapping, effectively avoiding the situation where traditional particle swarm optimization algorithms easily get trapped in local optima, thus improving global search capability and optimization efficiency. The resulting control strategy can achieve optimal output of distributed resources, minimize network losses and voltage deviations, thereby ensuring that the power distribution system can not only guarantee safe and stable operation when high-penetration distributed power sources are connected, but also achieve economical and efficient optimized scheduling, thus improving the overall performance of the system. This solves the technical problem in existing technologies where the lack of effective multi-objective coordination mechanisms and optimization algorithms makes it difficult to balance safe and economical operation in power distribution systems with high-penetration distributed power sources. Attached Figure Description

[0025] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0026] Figure 1 This is a schematic diagram of an optional multi-objective optimization method for power distribution system operation according to an embodiment of this application;

[0027] Figure 2 This is a flowchart of an optional vector machine model optimization using a target optimization algorithm according to an embodiment of this application;

[0028] Figure 3 This is a flowchart illustrating an optional multi-objective optimization method for power distribution system operation according to an embodiment of this application.

[0029] Figure 4 This is a schematic diagram of an optional multi-objective optimization device for power distribution system operation according to an embodiment of this application. Detailed Implementation

[0030] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0031] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0032] According to an embodiment of this application, a method embodiment of a multi-objective optimization method for power distribution system operation is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0033] It should be noted that the information collected in this application (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) are information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of this data all comply with relevant laws, regulations, and standards, necessary confidentiality measures have been taken, and they do not violate public order and good morals. Corresponding access points are provided for users to choose to authorize or refuse. For example, interfaces are set up between this system and relevant users or organizations, providing users with corresponding access points to choose to agree to or refuse automated decision-making results; if the user chooses to refuse, the process proceeds to the expert decision-making stage.

[0034] According to the embodiments of this application, an intelligent power distribution network system can be used as the execution subject of the multi-objective optimization method for power distribution system operation in the embodiments of this application. The system can be a software system or an embedded system combining software and hardware. Of course, the execution subject of the method in the embodiments of this application can also be other forms of execution subject, such as devices, equipment, etc. It should be known by those skilled in the art that this application does not particularly limit the specific form of the execution subject.

[0035] Figure 1 This is a schematic diagram of a multi-objective optimization method for power distribution system operation according to an embodiment of this application, such as... Figure 1 As shown, the method includes the following steps:

[0036] Step S101: The photovoltaic power output is predicted based on meteorological data using a photovoltaic power prediction model. The photovoltaic power prediction model is a vector machine model optimized by an objective optimization algorithm. The objective optimization algorithm is used to control the prediction accuracy of the vector machine model by adjusting the penalty parameter and the kernel parameter of the radial basis function.

[0037] Optionally, photovoltaic power generation, as an intermittent energy source, is significantly affected by weather conditions and other factors, exhibiting considerable randomness and volatility in its output power. The photovoltaic power prediction model in this embodiment can be constructed based on Support Vector Machines (SVM). This model utilizes historical meteorological data such as cloud cover, temperature, humidity, and solar radiation intensity as input features to predict the output power of photovoltaic power plants, facilitating accurate estimation of photovoltaic output power. Support Vector Machines are a supervised learning algorithm that effectively handles complex nonlinear relationships in photovoltaic power prediction by finding the optimal hyperplane for data partitioning or regression prediction. In regression tasks, the goal of a Support Vector Machine is to find a function that minimizes the deviation between the vast majority of training data points and the function, while ensuring the flatness of the function itself to improve the model's generalization ability. Since photovoltaic prediction essentially maps a continuous power output value to historical meteorological data and photovoltaic power plant operating parameters, this embodiment uses a Support Vector Machine for regression analysis to construct the photovoltaic power prediction model.

[0038] During the model training phase using historical data, vector opportunity learns the complex nonlinear relationship between input features and power output. This application specifically selects radial basis functions as kernel functions, which can transform the intricate nonlinear relationships in the original data into a linearly separable relationship by projecting them onto a higher-dimensional feature space, thus effectively capturing the complex nonlinear relationship between input features and power output.

[0039] However, the performance of vector machine models is highly dependent on their parameter settings, particularly the penalty parameter (C) and the kernel parameter (γ) of the radial basis functions (RBF). The penalty parameter controls the severity of the penalty applied to training data that exceeds the error tolerance range; an excessively large penalty parameter value may lead to overfitting, while an excessively small value may lead to underfitting, thus affecting prediction accuracy. Meanwhile, the kernel parameter of the RBF kernel itself determines the width of the influence range of a single training sample, directly affecting the shape and complexity of the decision boundary. Therefore, determining the optimal combination of the penalty parameter and the RBF kernel parameter is crucial for constructing a high-precision prediction model.

[0040] This application introduces an objective optimization algorithm to optimize the parameters of the vector machine model. For example, the Grey Wolf Optimization Algorithm (GWO) can be introduced. The Grey Wolf Optimization Algorithm is a metaheuristic optimization method that simulates the hunting behavior of a grey wolf pack. Its task is to find the optimal parameter combination that maximizes the accuracy of the photovoltaic power prediction model within the search space composed of the penalty parameter and the kernel parameters of the radial basis functions. This Grey Wolf Optimization Algorithm efficiently searches for the optimal solution in the parameter space by simulating the social hierarchy and cooperative hunting mechanism of a wolf pack, dynamically adjusting the penalty parameter and kernel parameter to achieve the best prediction accuracy. It has the advantages of avoiding local optima and fast convergence speed, and performs better than traditional optimization algorithms. The optimized vector machine model in this application can more accurately predict photovoltaic power output, providing reliable data support for the optimized scheduling of the power distribution system, thereby better addressing the high-penetration distributed power source access and facilitating the safe and economical operation of the system.

[0041] Step S102: Based on the photovoltaic power output predicted by the photovoltaic power prediction model, a multi-objective optimization model is constructed. The solution objectives of the multi-objective optimization model include the grid loss, voltage deviation and operating cost of the power distribution system.

[0042] Optionally, in distribution systems with high penetration of distributed power generation, focusing solely on single objectives such as maximizing photovoltaic (PV) consumption or minimizing operating costs often fails to comprehensively guarantee the system's safe, stable, and economical operation. Therefore, embodiments of this application can use grid losses, voltage deviation, and operating costs as the main solution objectives of a multi-objective optimization model. The grid loss objective is used to minimize the power loss of each branch in the distribution network, thereby improving system efficiency; the voltage deviation objective is used to ensure that the voltage at each node in the distribution network remains within a reasonable range, guaranteeing power quality; and the operating cost objective focuses on the overall economic benefits of the system, including the economic benefits and maintenance costs of distributed power generation, to achieve economical operation.

[0043] By incorporating the three objective factors of grid loss, voltage deviation, and operating cost of the distribution system into a multi-objective optimization model, a comprehensive balance can be struck between technical and economic factors in system operation. The embodiments of this application, through multi-objective optimization, not only consider the impact of distributed generation on grid loss and voltage stability but also take into account the system's economics, ensuring that the distribution system achieves an optimal balance between safety, stability, and economy in complex scenarios with high distributed generation penetration.

[0044] Step S103: Determine the hierarchical weights of each solution objective in the multi-objective optimization model using the analytic hierarchy process (AHP).

[0045] Optionally, when constructing a multi-objective optimization model, since different optimization objectives such as grid losses, voltage deviations, and operating costs are conflicting and have varying degrees of importance, the Analytic Hierarchy Process (AHP) can be used to determine the weights of each objective, thereby transforming the multi-objective problem into a solvable single-objective problem. The AHP is an effective decision analysis method that constructs a judgment matrix to compare each optimization objective pairwise and determine the relative importance of each solution objective. For example, each solution objective can first be compared with other objectives, assigned a relative importance score, then a judgment matrix can be constructed and its eigenvectors calculated. The components of the eigenvectors are used as the hierarchical weights of each objective. The hierarchical weights of each solution objective reflect the importance of each objective in the overall optimization, ensuring that the optimization results take into account the characteristics of each objective while achieving the overall optimal operation of the system.

[0046] Step S104: Convert the multi-objective optimization model into a single-objective solution function according to the hierarchical weights of each solution objective.

[0047] Optionally, in multi-objective optimization problems, different objectives often conflict and are difficult to optimize simultaneously. This application's embodiments can perform weighted processing based on the hierarchical weights of each optimization objective, facilitating the transformation of complex multi-objective optimization models into easier-to-solve single-objective problems. For example, the hierarchical weights of each solution objective determined by the Analytic Hierarchy Process (AHP) reflect the importance of each objective in the overall optimization. Multiplying the hierarchical weights of each objective by their corresponding objective functions and summing the weighted objective functions forms a comprehensive single-objective solution function. This facilitates unifying multiple previously dispersed objectives into a single comprehensive objective, allowing the optimization problem to be solved using standard single-objective optimization algorithms, thereby achieving overall optimal system operation while satisfying the relative importance of each objective.

[0048] Step S105: Solve the single-objective solution function according to the target particle swarm optimization algorithm to obtain the control strategy for the operation of the power distribution system. The target particle swarm optimization algorithm is a particle swarm optimization algorithm that uses chaotic mapping to determine the initial population.

[0049] Optionally, traditional particle swarm optimization (PSO) algorithms have limitations when solving optimization problems in continuous spaces, such as being prone to getting trapped in local optima. This limitation is particularly evident in power distribution network optimization problems. PSO algorithms can randomly assign particles, and the optimal solution is highly dependent on the initial particle distribution. The more uniform the initial particle distribution, the richer the diversity of the swarm, and the faster the optimal solution is obtained.

[0050] The target particle swarm optimization algorithm in this application is a particle swarm optimization algorithm that uses chaotic mapping to determine the initial population. Chaotic mapping can generate sequences with pseudo-randomness and ergodicity. Initializing the particle swarm positions through chaotic mapping can make the particles more evenly distributed in the solution space, avoiding the problem of uneven initial population distribution that may be caused by traditional random initialization, thereby enhancing the global search capability of the algorithm. This application embodiment introduces chaos theory to initialize the particle swarm algorithm, which is beneficial in the later stages of optimization when the convergence speed is slow and it is easy to get trapped in local optima, thus improving the performance and efficiency of the algorithm.

[0051] In an optional embodiment, before predicting photovoltaic power output based on meteorological data using a photovoltaic power prediction model, the method further includes: establishing an initial vector machine (API) model for the smart distribution network system, wherein the kernel function of the API model is a radial basis function (RBF), the kernel parameters of the RBF are used to control the model complexity, and the model parameters of the API model also include penalty parameters used to control the model's tolerance to training data errors. Then, a target optimization algorithm is used to automatically adjust the penalty parameters and kernel parameters in the API model to obtain a target vector machine (API) model. The target optimization algorithm is used to simulate the leadership structure and social behavior of a wolf pack, and through an iterative process, finds the optimal combination of penalty parameters and kernel parameters in the parameter space of the API model. In each iteration, the target optimization algorithm uses the directions indicated by the three leader wolves to guide the population update to approach the optimal solution, and the target optimization algorithm gradually reduces the search range through a linear reduction strategy until convergence. Finally, the target vector machine model is iteratively trained using a training dataset to obtain the photovoltaic power prediction model.

[0052] Optionally, when constructing a photovoltaic power prediction model, a smart distribution network system can use an Initial Vector Machine (IPM) model as its basic framework. This IPM model uses radial basis functions (RBFs) as its kernel function, and the kernel parameters of the RDFs control the model's complexity, ensuring that the model can adapt to the complex nonlinear relationships in photovoltaic power prediction. Simultaneously, the IPM model introduces a penalty parameter to adjust the model's tolerance to training data errors, thereby achieving a balance between fitting accuracy and generalization ability. However, the performance of the IPM model is highly dependent on the reasonable setting of the kernel and penalty parameters. Traditional manual adjustment methods often struggle to find the optimal parameter combination; therefore, an objective optimization algorithm is needed to automatically adjust the penalty parameters and the kernel parameters of the RDFs.

[0053] Optionally, embodiments of this application employ a goal-oriented optimization algorithm to optimize the parameters of the initial vector machine model. For example, the gray wolf optimization algorithm can be used, which simulates the leadership structure and social behavior of a wolf pack, searching for the optimal solution in the parameter space through an iterative process. In each iteration, the goal-oriented optimization algorithm uses the positional information of the three leader wolves (α, β, δ) to guide the population update, gradually approaching the optimal solution. Simultaneously, a linear reduction strategy is used to gradually narrow the search range, improving search accuracy until the algorithm converges. Finally, a goal-oriented vector machine model is constructed using the optimized parameters, and iteratively trained using a training dataset containing historical meteorological data and photovoltaic power output data, resulting in a high-precision photovoltaic power prediction model. This not only improves the model's prediction accuracy but also enhances its generalization ability, facilitating the optimized operation of smart distribution network systems.

[0054] For example, this application can use the gray wolf optimization algorithm to optimize the regularization parameter c and kernel function parameters of the initial vector machine model. The Grey Wolf Optimization Algorithm represents a novel metaheuristic technique for solving optimization problems, achieving good results. This algorithm mimics the leadership hierarchy and hunting mechanisms of grey wolves; for example, it can provide four types of grey wolves, namely... , , and These four types can be used to simulate leadership hierarchies, facilitating the simulation of leadership levels. Hunting, i.e., optimization, can be achieved by three wolves ( , and )guide.

[0055] During the hunt, gray wolves surround their prey. Mathematically, this can be represented by formulas (1) and (2):

[0056] Formula (1);

[0057] Formula (2);

[0058] in, Used to indicate the current iteration. and Used to represent the coefficient vector Used to represent the prey's position vector Used to represent the gray wolf's position vector.

[0059] and Vector calculations can be performed using formulas (3) and (4):

[0060] Formula (3);

[0061] Formula (4);

[0062] in, The component can decrease linearly from 2 to 0 during the iteration process. , Used to represent a random vector in [0,1].

[0063] To simulate the hunting process of a gray wolf, we assume... (Best candidate solution) and With a better understanding of the prey's possible location, the three best solutions obtained so far can be saved and forced to be used by other search agents (including...). Update the location based on the location of the best search agent. Refer to formulas (5), (6), and (7) to update the gray wolf's location:

[0064] Formula (5);

[0065] Formula (6);

[0066] Formula (7);

[0067] Figure 2 The flowchart illustrating the optimization of a vector machine model using a target optimization algorithm according to an embodiment of this application is shown. The process can begin with initialization, setting the number of search agents n, parameters a, and coefficient vectors. and Maximum number of iterations (Maxiter), upper bound of the variable and lower bound of variables The current iteration number is set to iter=0, an initial cost function is defined, and an initial population is randomly generated. Each search agent represents a set of parameters in the vector machine model, such as the penalty parameter C and the kernel parameter γ. The fitness of each agent is calculated. , and The fitness is evaluated by its prediction accuracy, which assesses its effectiveness in predicting photovoltaic power output. Fitness calculation is accomplished by inputting the training dataset, executing a vector machine model using k-fold cross-validation, and calculating the prediction accuracy to determine the prediction precision.

[0068] In each iteration, for each search agent, the position of the search agent can be updated using the positions of the three leader wolves (X1, X2, X3). This is a strategy that simulates the social behavior of a wolf pack, facilitating the guidance of the population towards the optimal solution. The position of the current search agent can be updated using formula (8):

[0069] Formula (8);

[0070] Furthermore, it updates its parameter 'a' and coefficient vector based on the current location of the search agent. and The fitness is then recalculated. The target vector machine model is iterated over the training dataset until the maximum number of iterations (Maxiter) is reached or the convergence condition is met. Finally, the optimal combination of penalty and kernel parameters found in the parameter space of the initial vector machine model after optimization can be used to construct a photovoltaic power prediction model to achieve higher accuracy predictions.

[0071] In one optional embodiment, a photovoltaic power prediction model is obtained by iteratively training the target vector machine model using a training dataset. This includes: the smart distribution network system acquiring historical meteorological data from photovoltaic power plants as training samples, wherein the historical meteorological data includes cloud cover information, temperature range, humidity changes, solar radiation intensity, and photovoltaic system characteristic parameters; and acquiring historical photovoltaic power output data from the photovoltaic power plants as training sample labels. Then, the training samples and training sample labels are used as a training dataset, and the target vector machine model is iteratively trained using this training dataset to obtain the photovoltaic power prediction model.

[0072] Optionally, when constructing a photovoltaic (PV) power prediction model, the smart distribution network system can first collect historical meteorological data from PV power plants as training samples. This historical meteorological data includes cloud cover information, temperature range, humidity changes, solar radiation intensity, and PV system characteristic parameters, comprehensively reflecting the characteristics of the PV power plant's operating environment and being a key factor affecting PV power output. Furthermore, the smart distribution network system also needs to obtain historical PV power output data corresponding to the PV power plants as training sample labels. These training sample labels directly reflect the actual power output of the PV power plants under specific meteorological conditions.

[0073] Optionally, the smart distribution network system combines the training samples and their labels into a training dataset for iterative training of the target vector machine model. By continuously adjusting the model parameters, it can better fit the relationship between inputs such as meteorological data and outputs such as photovoltaic power in historical data. After multiple iterations of training, the model can learn the complex mapping relationship between meteorological conditions and photovoltaic power, ultimately obtaining a high-precision photovoltaic power prediction model, which facilitates reliable photovoltaic power prediction for the smart distribution network system.

[0074] For example, in the embodiments of this application, a photovoltaic power prediction model can be used to complete the prediction task. The specific definition can be found in formula (9):

[0075] Formula (9);

[0076] in, Used to represent the weight vector perpendicular to the hyperplane. Used to represent offset. The parameter ξ represents the size of the training set, C represents the regularization parameter, and ξ represents the tolerance for classification error. Used to represent input vectors.

[0077] In this embodiment, the radial basis function is selected as the kernel function of the vector machine model, as shown in formula (10):

[0078] Formula (10);

[0079] in, The radial basis function, used to represent the output vector, involves parameters. ,parameter The value of varies between 1 and 100.

[0080] The training samples required for training the target vector machine model can be power generation data and meteorological information of a specific photovoltaic power station over a period of time. The training samples can include environmental features such as cloud cover, daily maximum temperature, daily minimum temperature, maximum humidity, and minimum humidity at the corresponding time point, as well as solar radiation intensity, photovoltaic array area, and photoelectric conversion efficiency. In addition, they can also include the corresponding power output variables at the corresponding time point. In this application, the mean absolute percentage error and root mean square error can be used to measure the model's predictive performance.

[0081] In one optional embodiment, a multi-objective optimization model is constructed based on the photovoltaic output power predicted by the photovoltaic power prediction model. This includes: the smart distribution network system constructing a grid loss objective function and a voltage deviation objective function based on the photovoltaic output power predicted by the photovoltaic power prediction model. The voltage deviation objective function minimizes the voltage deviation at each node in the distribution network, and the grid loss objective function minimizes the grid loss of each branch in the distribution network. An operating cost objective function is also constructed to minimize the overall operating cost of the distribution network. The harmonic result of the grid loss objective function, the voltage deviation objective function, and the operating cost objective function is then used as the multi-objective optimization model.

[0082] Optionally, when optimizing operation, the smart distribution network system can first use the photovoltaic power output predicted by the photovoltaic power prediction model as input to construct a grid loss objective function and a voltage deviation objective function. The grid loss objective function minimizes the grid loss of each branch in the distribution network. By optimizing the current distribution and power flow of each branch, it reduces energy loss during power transmission, thereby improving the overall efficiency of the system. The voltage deviation objective function minimizes the voltage deviation of each node in the distribution network, ensuring that the node voltage is within the specified range, guaranteeing power quality, and avoiding adverse effects on equipment and users due to excessively high or low voltage. The smart distribution network system also constructs an operating cost objective function, which minimizes the overall operating cost of the distribution network, including the generation cost of distributed resources, maintenance costs, and the economic costs of interacting with the grid.

[0083] Optionally, the objective functions of grid loss, voltage deviation, and operating cost can be comprehensively harmonized to form a multi-objective optimization model. This multi-objective optimization model can comprehensively consider the system's economy, security, and power quality, and achieve the optimal operating strategy for the distribution system under conditions of high-penetration distributed power source access by balancing the relationships between different objectives.

[0084] For example, the expression for the minimum voltage deviation of a node in a power distribution network can be found in formula (11):

[0085] Formula (11);

[0086] in, Used to represent nodes node voltage, Used to represent the minimum voltage deviation of nodes in a network.

[0087] In one alternative embodiment, the parameters of the power grid loss objective function include the resistance of each branch in the distribution network, the active and reactive power flowing through the branch, and the voltage of the branch terminal nodes.

[0088] Optionally, branch resistance characterizes the inherent properties of the line and directly affects energy loss when current flows through it. The higher the resistance, the greater the loss for the same current. The flow of active and reactive power determines the magnitude and phase of the current, thus affecting voltage drop and power loss in the branch. For example, active power transmission leads to heat loss in the line, while reactive power flow affects voltage levels and power factor, further impacting the overall system loss. Changes in voltage at branch terminal nodes reflect the power transmission status of the branch and its impact on loss calculations in subsequent branches.

[0089] Optionally, by adjusting branch current, optimizing voltage distribution, or improving power factor, grid losses can be effectively reduced, which not only improves the system's operating efficiency but also reduces energy waste during power transmission.

[0090] For example, the objective function for power grid losses can be found in formula (12):

[0091] Formula (12);

[0092] In the formula, Used to indicate a branch The resistance; Used to indicate flow The active power of the branch circuit; Used to indicate flow The reactive power of the branch circuit; Used to represent The node voltage of the branch terminal node; Used to indicate the number of branches.

[0093] In one optional embodiment, the parameters of the operating cost objective function include the economic benefits of distributed resource generation and the maintenance costs of distributed resource generation. The economic benefits of distributed resource generation are determined by the distributed resource generation revenue coefficient and the distributed resource output value. The maintenance costs of distributed resource generation are determined by the distributed resource maintenance and operation cost coefficient and the distributed resource output value. The distributed resources include at least a photovoltaic power station.

[0094] Optionally, the power generation revenue factor reflects the economic value of a unit of electricity generated, while the output value of distributed resources determines the actual power generation. The product of the distributed resource power generation revenue factor and the distributed resource output value quantifies the economic benefits of distributed resources such as photovoltaic power plants over a certain period of time. This benefit reflects the positive impact of distributed resources on the economic contribution of the system.

[0095] Optionally, the maintenance cost coefficient reflects the maintenance cost required per unit of power generation, while the output value determines the actual maintenance cost. Maintenance costs are an unavoidable economic expense during the operation of distributed resources and need to be controlled during the optimization process. By comprehensively considering economic benefits and maintenance costs, the operating cost objective function can comprehensively evaluate the economic viability of distributed resources such as photovoltaic power plants in the system, thereby achieving a balance between maximizing economic benefits and minimizing operating costs during the optimization process.

[0096] For example, the objective function expression for minimizing the operating cost of a distribution network can be found in formula (13):

[0097] Formula (13);

[0098] in:

[0099] Formula (14);

[0100] Formula (15);

[0101] In the formula, Used to represent the economic benefits of distributed resource generation; Used to represent the maintenance cost of distributed resource generation; Used to represent the revenue coefficient of distributed resource generation; A cost coefficient used to represent the maintenance and operation of distributed resources; Used to represent the output value of distributed resources.

[0102] In one optional embodiment, the hierarchical weights of each solution objective in the multi-objective optimization model are determined by the analytic hierarchy process (AHP). This includes: the smart distribution network system comparing each solution objective in the multi-objective optimization model pairwise with all other solution objectives to determine the relative importance score of that solution objective in the multi-objective optimization model; constructing a judgment matrix based on the relative importance scores of each solution objective in multiple objective optimization models; where each element in the judgment matrix represents the relative importance of a pair of solution objectives in the multi-objective optimization model; the rows and columns of the judgment matrix correspond to different solution objectives in the multi-objective optimization model; determining the eigenvectors of the judgment matrix; and then using each component of the eigenvectors as a hierarchical weight of a solution objective.

[0103] Optionally, when constructing a multi-objective optimization model, the smart distribution network system employs the analytic hierarchy process (AHP) to determine the weights of each optimization objective. First, the smart distribution network system can compare each objective pairwise with other objectives in the multi-objective optimization model, assessing their relative importance and assigning a relative importance score to each objective. This score reflects the priority of different objectives in the overall optimization problem. Based on this score, the smart distribution network system can construct a judgment matrix, where each element represents the relative importance of a pair of objectives. The rows and columns of the judgment matrix correspond to different optimization objectives, thus enabling the smart distribution network system to systematically quantify the relative importance relationships between each objective.

[0104] Optionally, the smart distribution network system determines the hierarchical weights of each solution objective by calculating the eigenvectors of the judgment matrix. Each component of the eigenvector corresponds to an optimization objective, and its value reflects the importance of that objective in the entire optimization model. These hierarchical weights enable the smart distribution network system to perform weighted processing according to the importance of each solution objective, thereby effectively transforming the complex multi-objective optimization problem into a single-objective optimization problem and ensuring that the optimization results better balance the relationships between different objectives.

[0105] In an optional embodiment, after treating each component of the feature vector as a hierarchical weight of a solution objective, the method further includes: the smart distribution network system detecting the consistency ratio among the hierarchical weights of all solution objectives in the multi-objective optimization model, wherein the consistency ratio is used to determine the degree of logical consistency between the construction of the judgment matrix and the allocation of hierarchical weights. If the consistency ratio is detected to be greater than a preset threshold, the judgment matrix is ​​adjusted until the consistency ratio is detected to be less than or equal to the preset threshold.

[0106] Optionally, when constructing a multi-objective optimization model using the analytic hierarchy process (AHP), the smart distribution network system calculates the hierarchical weights of each optimization objective and evaluates the logical rationality of these weights using a consistency ratio. The consistency ratio is a key indicator characterizing the logical consistency of the judgment matrix construction and hierarchical weight allocation. A high consistency ratio suggests potential logical contradictions or subjective biases in the relative importance scores among the solution objectives in the judgment matrix, which could affect the reliability of optimization decisions. Therefore, the smart distribution network system needs to check the consistency ratio to ensure the rationality of the weight allocation.

[0107] Optionally, when the consistency ratio is detected to be greater than a preset threshold, it indicates that the current judgment matrix may be unreasonable and needs to be adjusted. The adjustment process may involve reassessing the relative importance of each optimization objective, correcting the scores in the judgment matrix, until the consistency ratio drops below the preset threshold. In this way, the embodiments of this application ensure that the construction of the judgment matrix and the allocation of hierarchical weights are logically highly consistent, and can better balance the relationship between different solution objectives.

[0108] In one optional embodiment, the multi-objective optimization model is converted into a single-objective solution function according to the hierarchical weights of each solution objective. This includes: the smart distribution network system performs weighted calculations on the functions corresponding to each solution objective in the multi-objective optimization model according to the hierarchical weights of each solution objective to obtain a single-objective solution function.

[0109] Optionally, when performing multi-objective optimization, smart distribution network systems first need to transform the complex multi-objective problem into a more easily solvable single-objective problem. To this end, the smart distribution network system weights each objective function in the multi-objective optimization model based on the hierarchical weights of the optimization objectives determined by the Analytic Hierarchy Process (AHP). These objective functions include grid losses, voltage deviation, and operating costs, each reflecting an important aspect of the smart distribution network system's operation. By multiplying each objective function by its corresponding hierarchical weight, the smart distribution network system can quantify the importance of different objectives and incorporate them into a unified optimization framework. This weighting method not only considers the relative importance of each objective but also ensures a balance among the objectives during the optimization process.

[0110] Optionally, after completing the weighted calculation, the smart distribution network system sums all the weighted objective functions to obtain a comprehensive single-objective solution function. This single-objective function can integrate the characteristics of all optimization objectives and measure the overall performance of the system through a unified objective function. This transformation allows the originally complex multi-objective optimization problem to be solved using a standard single-objective optimization algorithm, thereby achieving the overall optimal operation of the distribution system while satisfying the relative importance of each solution objective, thus improving optimization efficiency.

[0111] For example, in this embodiment, the three objectives of minimizing distribution network losses, minimizing network voltage deviation rate, and minimizing distribution network operating costs can be harmonized as the final objective function. This embodiment can determine the weights of the three objectives separately and then harmonize them, successfully transforming a multi-objective optimization problem into a single-objective optimization problem, thus facilitating the optimal adjustment of distributed resources.

[0112] This application uses the analytic hierarchy process (AHP) to determine the importance of the three objectives. Used to represent the objective function For the target The degree of effect is determined by formula (16):

[0113] Formula (16);

[0114] Then, construct the corresponding judgment matrix based on the obtained element values. , refer to formula (17):

[0115] Formula (17);

[0116] In the formula, the matrix In This matrix is ​​used to represent the number of objective functions that need to be optimized. Using this matrix, we can leverage... The proportion of each objective was calculated. The corresponding geometric mean can be calculated using formula (18):

[0117] Formula (18);

[0118] Therefore, this application determines the weights of each objective by constructing a judgment matrix, and the calculation formula is referenced in formula (19):

[0119] Formula (19);

[0120] Finally, the single-objective function expression is obtained, as shown in formula (20):

[0121] Formula (20);

[0122] In an optional embodiment, the method further includes: during the process of converting a multi-objective optimization model into a single-objective solution function according to the hierarchical weights of each solution objective, the smart distribution network system determines the target constraints of the single-objective solution function based on all the constraints of the multi-objective optimization model, wherein the constraints of the multi-objective optimization model include: power flow equation constraints, node voltage magnitude constraints, branch thermal constraints, and distributed resource output regulation constraints.

[0123] Optionally, in the process of converting a multi-objective optimization model into a single-objective solution function, the smart distribution network system needs to consider not only the hierarchical weights of each optimization objective but also ensure that the optimization process conforms to various constraints in actual operation, facilitating the safe and stable operation of the distribution system. For example, the constraints of the multi-objective optimization model include power flow equation constraints to ensure power balance and reasonable transmission in the power system; node voltage amplitude constraints to ensure that the voltage at each node is within a specified safe range, avoiding damage to equipment and users due to excessively high or low voltage; branch thermal constraints to prevent thermal faults caused by line overload; and output regulation constraints for distributed resources to ensure that the output of distributed power sources is within a reasonable range, avoiding impact on the power grid. These constraints together constitute the objective constraints of the single-objective solution function, providing a clear boundary for the optimization process.

[0124] During the transformation process, the smart distribution network system incorporates these constraints into the construction of the single-objective solution function, ensuring that the optimization result not only achieves optimal economic and technical objectives but also meets the safety and reliability requirements of actual operation. In this way, the single-objective solution function can comprehensively consider all aspects of system operation during the optimization process, thereby achieving overall optimized operation of the distribution system. The embodiments of this application, through a method that comprehensively considers constraints, can effectively guide the actual operation and scheduling of the smart distribution network.

[0125] For example, power flow equation constraints, nodal voltage magnitude inequality constraints, and branch thermal constraints can be referenced in formula (21):

[0126] Formula (21);

[0127] in, Used to indicate the first Active power output of distributed road resources; Used to indicate the first Reactive power output of distributed resources in the road network; Used to indicate the first The active power of the road load; Used to indicate the first The reactive power of the road load. Used to represent the real part of the nodal admittance matrix. Used to represent the imaginary part of the nodal admittance matrix. Used to indicate except Another branch road outside.

[0128] Other relevant constraints can be found in formulas (22) and (23):

[0129] Formula (22);

[0130] Formula (23);

[0131] in, Used to represent Apparent power of the branch, Used to indicate flow The active power of the branch circuit; Used to indicate flow The reactive power of the branch circuit Used to represent The upper limit of apparent power of the branch.

[0132] Secondly, the output regulation constraints of distributed resources need to be considered. Both distributed and centralized distributed resources operate normally at rated power, which facilitates the full utilization of the active power generation of distributed resources. The regulation of the active power output of distributed resources is maintained or reduced, while the reactive power output is zero. This is because it is assumed by default that the distributed resource system operates near the unity power factor, as shown in formula (24).

[0133] Formula (24);

[0134] in, Used to represent The minimum active power output of the distributed resource; Used to represent The maximum active power output of the distributed resource.

[0135] Inverters can be controlled to provide or absorb reactive power. Furthermore, for reactive power control of distributed resources, a specific power factor can be provided to the reactive power via a power converter, based on the adjustment scheme of the control / dispatch center.

[0136] In high-ratio distributed resource integration scenarios, to prevent the compression ratio of a single distributed resource from exceeding its own adjustment capacity, a consistency reduction constraint is proposed. The reserve rate of a distributed resource refers to the ratio of the actual active power reduction value to the maximum active power value that can be generated. Consistency reduction ensures that the reserve rate of each distributed resource maintains a similar ratio, and its mathematical definition is given by formula (25):

[0137] Formula (25);

[0138] in, Used to represent optimized nodes The reduction value of active power output of distributed resources. Used to represent the maximum availability value of distributed resources. The setting used to indicate consistency reduction.

[0139] In one optional embodiment, the single-objective solution function is solved using a target particle swarm optimization algorithm to obtain the control strategy for the operation of the power distribution system. This includes: the smart power distribution system generates a particle swarm at initial positions based on a chaotic sequence; using the logistic mapping in chaos theory, a series of chaotic numbers are generated; the logistic mapping generates a pseudo-random sequence through nonlinear iteration, which is used to break the regularity of the particle swarm positions. The generated chaotic numbers are then used to initialize the positions of the particle swarm. After initialization, the target particle swarm optimization algorithm searches for the optimal solution in the solution space of the single-objective solution function, where each particle represents a potential solution. Then, based on the optimal solution found in the solution space of the single-objective solution function, the control strategy for the operation of the power distribution system is determined. This control strategy includes: the optimal output value of distributed resources, the optimal network loss for the power distribution system operation, and voltage deviation.

[0140] Optionally, the smart distribution network system employs a particle swarm optimization algorithm based on chaos theory during the optimization process to improve optimization efficiency and global search capability. The smart distribution network system utilizes the logistic mapping in chaos theory to generate a series of chaotic numbers. The logistic mapping generates a pseudo-random sequence through nonlinear iteration, which possesses good ergodicity and irregularity, effectively breaking the regular distribution of particle initial positions in traditional particle swarm algorithms. The smart distribution network system initializes the positions of the particle swarm based on these chaotic numbers, making the particle distribution in the solution space more uniform, thereby enhancing the algorithm's global search capability and avoiding getting trapped in local optima. After initialization, the target particle swarm optimization algorithm begins iterative search in the solution space of the single-objective solution function, where each particle represents a potential solution, continuously updating its position and velocity to find the optimal solution.

[0141] During the optimization process, the particle swarm optimization algorithm gradually approximates the global optimum by evaluating the fitness value of each particle. Ultimately, based on the optimal solution found in the solution space of the single-objective solution function, the smart distribution grid system determines the control strategy for the distribution system. This control strategy includes the optimal output value of distributed resources, the optimal network losses for distribution system operation, and voltage deviation. In this way, the smart distribution grid system can achieve the optimal balance between system security, stability, and economy even with high penetration of distributed power sources.

[0142] For example, this application embodiment utilizes chaotic mapping to extract the initial population of the particle swarm optimization algorithm, ensuring that particles are uniformly distributed in the solution space. Chaos refers to nonlinear motion that can traverse all situations within a specified range. A chaotic sequence is used to represent all states in a specified space and can be generated through mapping. Chaotic mapping has unpredictable characteristics, but certain laws can still be used for prediction during the mapping process. The chaotic mapping method is logical mapping, referencing formula (26):

[0143] Formula (26);

[0144] in, , Used to indicate passage of The value obtained from the second Logistic mapping. Used to represent control variables. When At that time, the system is in a completely chaotic state, and the range of the chaotic space is... .

[0145] The steps of using chaos to initialize a target particle swarm optimization algorithm may include: first, randomly generating... dimensional vector Furthermore, formula (25) is used to map other vectors, thereby generating chaotic sequences. Then the chaotic sequence From chaotic space Inverse mapping to the space of the optimal solution The particle position is , where i represents the i-th particle, N represents the particle swarm size, j represents the j-th dimension, and M represents the dimension of the space.

[0146] In an optional embodiment, the method further includes: during the process of the smart distribution network system searching for the optimal solution in the solution space of the single-objective solution function using the target particle swarm optimization algorithm, determining the fitness value of the solution represented by each particle, wherein the fitness of the solution characterizes the degree to which the solution satisfies the single-objective optimization function, and the lower the fitness value of the solution, the closer the solution is to the optimization objective of the objective function, and the greater the probability that the solution is the optimal solution.

[0147] Optionally, in the optimization process of a smart distribution network system, the objective particle swarm optimization algorithm can be used to find the optimal solution in the solution space of a single-objective solution function. Each particle in the solution space represents a potential solution, and the fitness value measures the quality of these solutions. The fitness value is determined by evaluating the single-objective optimization function and reflects the degree to which the current solution satisfies the optimization objective. For example, the lower the fitness value, the better the solution performs on the optimization objective and the closer it is to the optimal value of the objective function. The optimization objective can be minimizing network losses, voltage deviations, or operating costs, etc. The fitness value is not only used to judge the quality of solutions during the optimization process, but also to guide the particle swarm search direction.

[0148] For example, particles with high fitness values ​​can be identified as the initial particles of the population. When the particle swarm optimization algorithm gets stuck in a local optimum, it can select the historical optimal value of a particle during the iteration process and convert it into a chaotic sequence through inverse mapping to obtain the optimal position of the particle. Then, it can randomly replace the position of a particle in the current search space, thereby allowing the algorithm to escape the local optimum. Whether a particle gets stuck in premature convergence depends on the variance of the population fitness, and the calculation method is shown in formula (27).

[0149] Formula (27);

[0150] Where N represents the population size. Used to represent the fitness value of the i-th particle. Used to represent the average fitness of the current swarm of particles. Population variance. Used to reflect the premature maturation state of particles. When When the particle size falls below a certain threshold, the particle algorithm is found to be prematurely convergent. Chaos is then applied to handle the optimal particle, increasing population diversity.

[0151] In an optional embodiment, the method further includes: during the process of the smart distribution network system searching for the optimal solution in the solution space of a single-objective solution function using a target particle swarm optimization algorithm, at each iteration of the particle swarm, the fitness value of each particle in the particle swarm is calculated based on its position in the solution space, and the fitness value of each particle in the particle swarm is statistically analyzed, and the average fitness value of all particles is calculated. Then, based on the fitness value of each particle in the particle swarm and the average fitness value of all particles, the variance corresponding to the fitness value of the particle swarm is calculated, where the variance reflects the dispersion of the fitness value of the particle swarm. If the variance is detected to be less than a target threshold, new particle positions are generated using the logistic mapping in chaos theory to replace some particles in the particle swarm; if the variance is detected to be greater than or equal to the target threshold, the next iteration of the particle swarm continues until the number of iterations exceeds a preset number or the target particle swarm optimization algorithm is detected to have entered a convergent state.

[0152] Optionally, in the optimization process of the smart distribution network system, the target particle swarm optimization algorithm searches for the optimal solution of the single-objective solution function through iterative search. In each iteration, the smart distribution network system first calculates the fitness value of each particle in the solution space, which reflects the degree to which the solution represented by the particle satisfies the optimization objective. Subsequently, the smart distribution network system counts the fitness values ​​of all particles and calculates their average value. Based on the fitness value of each particle and the average fitness value, the variance of the particle swarm fitness value is further calculated. The variance is used to measure the dispersion of the particle fitness values, i.e., the diversity of the particle swarm. When the variance is less than the set target threshold, it indicates that the diversity of the particle swarm is low and it may be trapped in a local optimum. At this time, the smart distribution network system introduces the logistic mapping from chaos theory to generate new particle positions to replace some particles, thereby increasing the diversity of the particle swarm and helping the algorithm escape local optima.

[0153] Conversely, if the variance is greater than or equal to the target threshold, it indicates that the diversity of the particle swarm is still high, and the search process is still effective. Therefore, the smart distribution network system will continue with the next iteration. This process will continue until the number of iterations reaches a preset upper limit, or the target particle swarm optimization algorithm detects a convergence state, i.e., the fitness value no longer improves significantly. Through a dynamic adjustment mechanism, the smart distribution network system can effectively balance global search capabilities and local search capabilities, improve optimization efficiency, and ultimately determine the optimal control strategy that meets the system's operational requirements.

[0154] For example, detailed steps may include:

[0155] Step a: Select the optimal position during the iteration process and map it to the chaotic space [0, 1] using a logical function.

[0156] Step b: Generate a new sequence using logical variables and backmap the sequence to the population.

[0157] Step c: Calculate the optimal fitness value of the particle, determine whether the particle has escaped a local optimum, record the optimal fitness value, and set the corresponding particle as... , and .

[0158] Step d: Use the currently best chaotic particle to manage the particles in the particle swarm so that the particles leave the local optimum.

[0159] After performing chaotic initialization on the particle swarm, the particles are more evenly distributed in the search space, and the chaotic sequence can be used to reduce premature convergence, increase particle diversity, and improve the convergence speed of the algorithm.

[0160] For example, Figure 3 A flowchart illustrating the implementation of a multi-objective optimization method for power distribution system operation is provided. This application's embodiments improve the operating efficiency and economy of power distribution systems through photovoltaic power prediction models and optimization algorithms. The flowchart includes three main steps: power output prediction, optimization modeling, and model solving.

[0161] In the process of power output prediction, a photovoltaic power output prediction model can be established first, and then the model parameters can be optimized using the Grey Wolf optimization algorithm to improve the accuracy of the prediction model. That is, before predicting photovoltaic power output, an initial vector machine model is established, and the penalty parameters and kernel parameters in the model are automatically adjusted through the objective optimization algorithm.

[0162] In the process of optimizing the model, the objective function can be determined in sequence, optimization constraints with equal reserve rate can be established, and the objective function can be harmonized and normalized. That is, a multi-objective optimization model can be constructed, including the grid loss objective function, voltage deviation objective function and operating cost objective function, and the three can be harmonized into a multi-objective optimization model.

[0163] During the model solution process, the following steps can be performed sequentially: parameter initialization, generation of an initial chaotic sequence, mapping of the chaotic sequence to the solution space to obtain an initial particle swarm, calculation of the fitness of each particle, updating the optimal position of each individual particle and the global optimal position, and outputting the global optimal solution when the number of iterations or the result converges. In other words, the single-objective solution function is solved based on the objective particle swarm optimization algorithm to obtain the control strategy for the power distribution system. Specifically, a particle swarm optimization algorithm that determines the initial population can be used via chaotic mapping.

[0164] This application first establishes a photovoltaic power prediction model based on the gray wolf optimization algorithm and vector machines to accurately estimate the output power of photovoltaic power generation, providing a reliable basis for system scheduling. Second, considering network losses, voltage deviations, and operational economics, a multi-objective optimization function is constructed, and a consistency reduction constraint is introduced to ensure the fairness of distributed resource scheduling. Furthermore, the analytic hierarchy process (AHP) is used to transform the multi-objective optimization problem into a solvable single-objective problem. Finally, an improved particle swarm optimization algorithm incorporating chaotic initialization is employed to efficiently solve the power coordination optimization model of the distribution network, thereby improving overall control performance and economic benefits while ensuring the safe and stable operation of the system.

[0165] See Figure 4 According to another aspect of the embodiments of this application, a multi-objective optimization device for power distribution system operation is also provided, including: a power prediction unit 401, a model building unit 402, a hierarchical weight determination unit 403, a model conversion unit 404, and a control strategy determination unit 405.

[0166] The system includes the following components: a power prediction unit 401, which predicts photovoltaic power output based on meteorological data using a photovoltaic power prediction model (which is a vector machine model optimized by an objective optimization algorithm; the objective optimization algorithm controls the prediction accuracy of the vector machine model by adjusting the penalty parameter and the kernel parameter of the radial basis function); a model construction unit 402, which constructs a multi-objective optimization model based on the photovoltaic power output predicted by the photovoltaic power prediction model; the solution objectives of the multi-objective optimization model include grid losses, voltage deviation, and operating costs of the distribution system; a hierarchy weight determination unit 403, which determines the hierarchy weights of each solution objective of the multi-objective optimization model using the analytic hierarchy process (AHP); a model conversion unit 404, which converts the multi-objective optimization model into a single-objective solution function based on the hierarchy weights of each solution objective; and a control strategy determination unit 405, which solves the single-objective solution function using a target particle swarm optimization algorithm to obtain the control strategy for the operation of the distribution system; the target particle swarm optimization algorithm is a particle swarm optimization algorithm that uses chaotic mapping to determine the initial population.

[0167] Optionally, the multi-objective optimization device for power distribution system operation further includes: an initial model establishment unit for establishing an initial vector machine model, wherein the kernel function of the initial vector machine model is a radial basis function, the kernel parameters of the radial basis function are used to control the model complexity, and the model parameters of the initial vector machine model also include penalty parameters used to control the model's tolerance to training data errors; an objective model determination unit for automatically adjusting the penalty parameters and kernel parameters in the initial vector machine model through an objective optimization algorithm to obtain an objective vector machine model, wherein the objective optimization algorithm is used to find the optimal combination of penalty parameters and kernel parameters in the parameter space of the initial vector machine model through an iterative process by simulating the leadership structure and social behavior of a wolf pack; the objective optimization algorithm uses the directions indicated by the three leader wolves to guide the population update in each iteration to approach the optimal solution, and the objective optimization algorithm gradually reduces the search range through a linear reduction strategy until convergence; and a prediction model determination unit for iteratively training the objective vector machine model using a training dataset to obtain a photovoltaic power prediction model.

[0168] Optionally, the prediction model determination unit includes: a training sample acquisition subunit, used to acquire historical meteorological data of the photovoltaic power station as training samples, wherein the historical meteorological data includes cloud cover information, temperature range, humidity changes, solar radiation intensity, and photovoltaic system characteristic parameters; a training sample label acquisition subunit, used to acquire historical photovoltaic power output data of the photovoltaic power station as training sample labels; and an iterative training processing subunit, used to use the training samples and training sample labels as a training dataset, and use the training dataset to iteratively train the target vector machine model to obtain the photovoltaic power prediction model.

[0169] Optionally, the model building unit 402 includes: an objective function building subunit, used to build a grid loss objective function and a voltage deviation objective function based on the photovoltaic output power predicted by the photovoltaic power prediction model, wherein the voltage deviation objective function is used to minimize the voltage deviation of each node in the distribution network, and the grid loss objective function is used to minimize the grid loss of each branch in the distribution network; a cost function building subunit, used to build an operating cost objective function, wherein the operating cost objective function is used to minimize the overall operating cost of the distribution network; and an optimization model determination subunit, used to use the harmonic result of the grid loss objective function, the voltage deviation objective function, and the operating cost objective function as a multi-objective optimization model.

[0170] Optionally, the hierarchical weight determination unit 403 includes: a score determination subunit, used to determine the relative importance score of each solution objective in the multi-objective optimization model by comparing each solution objective with the other solution objectives in the multi-objective optimization model pairwise; a judgment matrix construction subunit, used to construct a judgment matrix based on the relative importance scores of each solution objective in multiple objective optimization models, wherein each element in the judgment matrix represents the relative importance of a pair of solution objectives in the multi-objective optimization model, and the rows and columns of the judgment matrix correspond to different solution objectives in the multi-objective optimization model; an eigenvector determination subunit, used to determine the eigenvectors of the judgment matrix; and a hierarchical weight determination subunit, used to treat each component of the eigenvector as a hierarchical weight of a solution objective.

[0171] Optionally, the multi-objective optimization device for power distribution system operation further includes: a consistency ratio detection unit, used to detect the consistency ratio among the hierarchical weights of all solution objectives in the multi-objective optimization model, wherein the consistency ratio is used to determine the degree of logical consistency between the construction of the judgment matrix and the allocation of hierarchical weights; and a judgment matrix adjustment unit, used to adjust the judgment matrix when the consistency ratio is detected to be greater than a preset threshold, until the consistency ratio is detected to be less than or equal to the preset threshold.

[0172] Optionally, the model transformation unit 404 includes: a solution function determination subunit, used to perform weighted calculations on the functions corresponding to each solution objective in the multi-objective optimization model according to the hierarchical weights of each solution objective, to obtain a single-objective solution function.

[0173] Optionally, the multi-objective optimization device for power distribution system operation further includes: an objective constraint determination unit, which determines the objective constraints of the single-objective solution function based on all the constraints of the multi-objective optimization model during the process of converting the multi-objective optimization model into a single-objective solution function according to the hierarchical weights of each solution objective. The constraints of the multi-objective optimization model include: power flow equation constraints, node voltage magnitude constraints, branch thermal constraints, and output regulation constraints of distributed resources.

[0174] Optionally, the control strategy determination unit 405 includes: a particle swarm generation subunit for generating a particle swarm at an initial position based on a chaotic sequence; a chaotic number generation subunit for generating a series of chaotic numbers using the logistic mapping in chaos theory, wherein the logistic mapping generates a pseudo-random sequence through nonlinear iteration, which is used to break the regularity of the particle swarm positions; a particle swarm position determination subunit for initializing the positions of the particle swarm using the generated chaotic numbers; and an optimal solution determination subunit for finding the optimal solution in the solution space of a single-objective solution function after initializing the positions of the particle swarm, using a target particle swarm optimization algorithm, wherein each particle represents a potential solution. The control strategy determination subunit is used to determine the control strategy for the operation of the power distribution system based on the optimal solution found in the solution space of the single-objective solution function, wherein the control strategy includes: the optimal output value of distributed resources, the optimal network loss and voltage deviation for the operation of the power distribution system.

[0175] Optionally, the multi-objective optimization device for power distribution system operation further includes: a fitness value determination unit, used to determine the fitness value of the solution represented by each particle during the process of finding the optimal solution in the solution space of the single-objective solution function through the objective particle swarm optimization algorithm. The fitness of the solution characterizes the degree to which the solution satisfies the single-objective optimization function. The lower the fitness value of the solution, the closer the solution is to the optimization objective of the objective function, and the greater the probability that the solution is the optimal solution.

[0176] Optionally, the multi-objective optimization device for power distribution system operation further includes: a fitness value calculation unit, used to calculate the fitness value of each particle in the particle swarm based on its position in the solution space during each iteration of the single-objective solution function, in the process of finding the optimal solution through the target particle swarm optimization algorithm; an average fitness value determination unit, used to count the fitness value of each particle in the particle swarm and calculate the average fitness value of all particles; a variance determination unit, used to calculate the variance corresponding to the fitness value of the particle swarm based on the fitness value of each particle in the particle swarm and the average fitness value of all particles, wherein the variance reflects the dispersion of the fitness value of the particle swarm; a particle position processing unit, used to generate new particle positions using the logistic mapping in chaos theory to replace some particles in the particle swarm when the variance is detected to be less than the target threshold; and a particle swarm iteration processing unit, used to continue to the next iteration of the particle swarm when the variance is detected to be greater than or equal to the target threshold, until the number of iterations is greater than a preset number or the target particle swarm optimization algorithm is detected to have entered a convergence state.

[0177] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, which stores a computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located performs the above-described multi-objective optimization method for power distribution system operation.

[0178] According to another aspect of the embodiments of this application, an electronic device is also provided, including one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by one or more processors, the one or more processors cause the one or more processors to perform the above-described multi-objective optimization method for power distribution system operation.

[0179] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program or instructions, which, when executed by a processor, implement the above-described multi-objective optimization method for power distribution system operation.

[0180] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0181] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0182] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0183] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0184] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0185] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0186] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A multi-objective optimization method for power distribution system operation, characterized in that, include: The photovoltaic power output is predicted based on meteorological data using a photovoltaic power prediction model. The photovoltaic power prediction model is a vector machine model optimized by an objective optimization algorithm. The objective optimization algorithm is used to control the prediction accuracy of the vector machine model by adjusting the penalty parameter and the kernel parameter of the radial basis function. Based on the photovoltaic power output predicted by the photovoltaic power prediction model, a multi-objective optimization model is constructed. The solution objectives of the multi-objective optimization model include grid loss, voltage deviation and operating cost of the power distribution system. The hierarchical weights of each solution objective in the multi-objective optimization model are determined using the analytic hierarchy process (AHP). The multi-objective optimization model is converted into a single-objective solution function based on the hierarchical weights of each solution objective; The single-objective solution function is solved using a target particle swarm optimization algorithm to obtain the control strategy for the operation of the power distribution system. The target particle swarm optimization algorithm is a particle swarm optimization algorithm that uses chaotic mapping to determine the initial population.

2. The multi-objective optimization method for power distribution system operation according to claim 1, characterized in that, Before predicting photovoltaic output power based on meteorological data using a photovoltaic power prediction model, the method further includes: An initial vector machine (API) model is established, wherein the kernel function of the API model is a radial basis function (RBF), the kernel parameters of the RBF are used to control the model complexity, and the model parameters of the API model also include penalty parameters used to control the model's tolerance to training data errors. The objective optimization algorithm automatically adjusts the penalty parameters and kernel parameters in the initial vector machine model to obtain the objective vector machine model. The objective optimization algorithm simulates the leadership structure and social behavior of a wolf pack and iteratively searches for the optimal combination of penalty parameters and kernel parameters in the parameter space of the initial vector machine model. In each iteration, the objective optimization algorithm uses the directions indicated by the three leader wolves to guide the population update to approach the optimal solution. Furthermore, the objective optimization algorithm gradually reduces the search range through a linear reduction strategy until convergence. The photovoltaic power prediction model is obtained by iteratively training the target vector machine model using the training dataset.

3. The multi-objective optimization method for power distribution system operation according to claim 2, characterized in that, The photovoltaic power prediction model is obtained by iteratively training the target vector machine model using a training dataset, including: Historical meteorological data of photovoltaic power plants are obtained as training samples. The historical meteorological data includes cloud cover information, temperature range, humidity changes, solar radiation intensity, and photovoltaic system characteristic parameters. Historical photovoltaic power output data of photovoltaic power plants are used as training sample labels; The training samples and their labels are used as the training dataset. The target vector machine model is then iteratively trained using the training dataset to obtain the photovoltaic power prediction model.

4. The multi-objective optimization method for power distribution system operation according to claim 1, characterized in that, Based on the photovoltaic power output predicted by the photovoltaic power prediction model, a multi-objective optimization model is constructed, including: Based on the photovoltaic power output predicted by the photovoltaic power prediction model, a grid loss objective function and a voltage deviation objective function are constructed. The voltage deviation objective function is used to minimize the voltage deviation of each node in the distribution network, and the grid loss objective function is used to minimize the grid loss of each branch in the distribution network. Construct an operating cost objective function, wherein the operating cost objective function is used to minimize the overall operating cost of the power distribution network; The harmonic result of the grid loss objective function, the voltage deviation objective function, and the operating cost objective function is used as the multi-objective optimization model.

5. The multi-objective optimization method for power distribution system operation according to claim 4, characterized in that, The parameters of the power grid loss objective function include the resistance of each branch in the distribution network, the active power and reactive power flowing through the branch, and the voltage of the branch terminal node.

6. The multi-objective optimization method for power distribution system operation according to claim 4, characterized in that, The parameters of the operating cost objective function include the economic benefits of distributed resource generation and the maintenance costs of distributed resource generation. The economic benefits of distributed resource generation are determined by the distributed resource generation revenue coefficient and the distributed resource output value. The maintenance costs of distributed resource generation are determined by the distributed resource maintenance and operation cost coefficient and the distributed resource output value. The distributed resources include at least photovoltaic power plants.

7. The multi-objective optimization method for power distribution system operation according to claim 1, characterized in that, The hierarchical weights of each solution objective in the multi-objective optimization model are determined using the analytic hierarchy process (AHP), including: The relative importance score of each solution objective in the multi-objective optimization model is determined by comparing each solution objective with the other solution objectives in pairs. A judgment matrix is ​​constructed based on the relative importance score of each solution objective in the multiple objective optimization models, wherein each element in the judgment matrix represents the relative importance of a pair of solution objectives in the multi-objective optimization model, and the rows and columns of the judgment matrix correspond to different solution objectives in the multi-objective optimization model, respectively. Determine the eigenvectors of the judgment matrix; Each component of the feature vector is used as a hierarchical weight for solving the objective.

8. The multi-objective optimization method for power distribution system operation according to claim 7, characterized in that, After treating each component of the feature vector as a hierarchical weight for solving the objective, the method further includes: The consistency ratio among the hierarchical weights of all the solution objectives in the multi-objective optimization model is detected, wherein the consistency ratio is used to determine the degree of logical consistency between the construction of the judgment matrix and the allocation of the hierarchical weights. If the consistency ratio is detected to be greater than a preset threshold, the judgment matrix is ​​adjusted until the consistency ratio is detected to be less than or equal to the preset threshold.

9. The multi-objective optimization method for power distribution system operation according to claim 1, characterized in that, The multi-objective optimization model is converted into a single-objective solution function based on the hierarchical weights of each solution objective, including: Based on the hierarchical weights of each solution objective, the functions corresponding to each solution objective in the multi-objective optimization model are weighted and calculated to obtain the single-objective solution function.

10. The multi-objective optimization method for power distribution system operation according to claim 1, characterized in that, The method further includes: In the process of converting the multi-objective optimization model into a single-objective solution function according to the hierarchical weights of each solution objective, the objective constraints of the single-objective solution function are determined based on all the constraints of the multi-objective optimization model. The constraints of the multi-objective optimization model include: power flow equation constraints, node voltage magnitude constraints, branch thermal constraints, and distributed resource output regulation constraints.

11. The multi-objective optimization method for power distribution system operation according to claim 1, characterized in that, The single-objective solution function is solved using the objective particle swarm optimization algorithm to obtain the control strategy for the operation of the power distribution system, including: Particle swarms are generated at the initial position based on chaotic sequences; Using the logistic mapping in chaos theory, a series of chaotic numbers are generated, wherein the logistic mapping generates a pseudo-random sequence through nonlinear iteration, which is used to break the regularity of particle swarm positions; The positions of the particle swarm are initialized using the generated chaos number; After initializing the positions of the particle swarm, the optimal solution is found in the solution space of the single-objective solution function using the target particle swarm optimization algorithm, where each particle represents a potential solution. Based on the optimal solution found in the solution space of the single-objective solution function, the control strategy for the operation of the power distribution system is determined, wherein the control strategy includes: the optimal output value of distributed resources, the optimal network loss and voltage deviation for the operation of the power distribution system.

12. The multi-objective optimization method for power distribution system operation according to claim 11, characterized in that, The method further includes: In the process of finding the optimal solution in the solution space of the single-objective solution function using the target particle swarm optimization algorithm, the fitness value of the solution represented by each particle is determined. The fitness value of the solution characterizes the degree to which the solution satisfies the single-objective optimization function. The lower the fitness value of the solution, the closer the solution is to the optimization objective of the objective function, and the greater the probability that the solution is the optimal solution.

13. The multi-objective optimization method for power distribution system operation according to claim 12, characterized in that, The method further includes: In the process of finding the optimal solution in the solution space of the single-objective solution function using the target particle swarm optimization algorithm, at each iteration of the particle swarm, the fitness value of each particle in the particle swarm is calculated according to its position in the solution space. The fitness value of each particle in the particle swarm is calculated, and the average fitness value of all particles is calculated. Based on the fitness value of each particle in the particle swarm and the average fitness value of all particles, the variance corresponding to the fitness value of the particle swarm is calculated, wherein the variance reflects the degree of dispersion of the fitness value of the particle swarm. If the variance is detected to be less than the target threshold, a new particle position is generated using the logistic mapping in chaos theory to replace a portion of the particles in the particle swarm. If the variance is detected to be greater than or equal to the target threshold, the next iteration of the particle swarm continues until the number of iterations exceeds the preset number or the target particle swarm optimization algorithm is detected to have entered a convergence state.

14. A multi-objective optimization device for power distribution system operation, characterized in that, include: A power prediction unit is used to predict photovoltaic power output based on meteorological data using a photovoltaic power prediction model. The photovoltaic power prediction model is a vector machine model optimized by an objective optimization algorithm. The objective optimization algorithm is used to control the prediction accuracy of the vector machine model by adjusting the penalty parameter and the kernel parameter of the radial basis function. The model building unit is used to build a multi-objective optimization model based on the photovoltaic power output predicted by the photovoltaic power prediction model. The solution objectives of the multi-objective optimization model include grid loss, voltage deviation and operating cost of the power distribution system. The hierarchical weight determination unit is used to determine the hierarchical weights of each solution objective in the multi-objective optimization model using the analytic hierarchy process (AHP). The model conversion unit is used to convert the multi-objective optimization model into a single-objective solution function according to the hierarchical weights of each solution objective; The control strategy determination unit is used to solve the single-objective solution function according to the target particle swarm optimization algorithm to obtain the control strategy for the operation of the power distribution system. The target particle swarm optimization algorithm is a particle swarm optimization algorithm that uses chaotic mapping to determine the initial population.

15. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein when the computer program is executed, the device containing the computer-readable storage medium performs the multi-objective optimization method for power distribution system operation as described in any one of claims 1 to 13.

16. An electronic device, characterized in that, It includes one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to perform the multi-objective optimization method for power distribution system operation as described in any one of claims 1 to 13.

17. A computer program product, characterized in that, It includes a computer program or instructions that, when executed by a processor, implement the multi-objective optimization method for power distribution system operation as described in any one of claims 1 to 13.

Images