Power distribution network reconstruction method and system

Abstract

The power distribution network reconstruction method comprises the steps that operation parameters of different wind power plants in a target power distribution network are acquired, a cumulative distribution function of each wind power plant is established according to the operation parameters in combination with a non-parameter estimation method, a power output scene of a first wind power plant is generated, and the operation parameters at least comprise historical wind speed and output power; reducing the generated power output scene of the first wind power plant based on a greedy strategy and a K-means algorithm to obtain a power output scene of a second wind power plant; establishing a power distribution network reconstruction model about the power output scene of the second wind power plant; and solving the power distribution network reconstruction model in combination with an improved multi-target Bayesian optimization algorithm. According to the method, the problems of the wind power absorption rate, the voltage stability and the system operation cost are considered, and the voltage stability can be guaranteed under the condition that the high wind power absorption rate can be guaranteed. According to the method, the spatial correlation of the output power of the multiple wind power plants is accurately represented, and the effectiveness of scene generation is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution network reconstruction, and in particular to a power distribution network reconstruction method and system. Background Art

[0002] Distribution network reconstruction plays a very important role in distribution networks. Reconstruction optimization is conducive to reducing power loss and improving efficiency. In recent decades, distributed power generation equipment such as wind turbines and solar panels have been widely used in distributed power grids, which not only saves resources but also protects the environment. Due to the relatively mature new energy access technology, more and more distributed energy sources are connected to distribution networks, which improves the operating efficiency of power systems and greatly reduces the operating costs of power systems. However, the uncertainty of new energy has brought challenges to the reliable operation of power systems. In the reconstruction of distribution networks containing distributed energy, uncertainty must be taken into account in the model. Therefore, the problem of distribution network reconstruction containing distributed energy has become a research focus.

[0003] The high penetration rate of renewable energy will lead to increased voltage instability in the system and affect the normal operation of the distribution network. Therefore, while absorbing more wind power, it is also necessary to ensure the voltage stability of the system. In existing research, some scholars prevent system overvoltage by limiting power, while others establish a model based on scenario generation and use particle swarm algorithm to optimize network topology and improve voltage stability. But in general, there are still problems such as difficulty in achieving both wind power absorption rate and voltage stability, and model calculation results deviating from the optimal solution. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides a distribution network reconstruction method and system, which can solve the problems mentioned in the background technology.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, the present invention provides a distribution network reconstruction method, comprising:

[0009] Acquire operating parameters of different wind farms in the target distribution network, establish a cumulative distribution function of each wind farm based on the operating parameters in combination with a non-parametric estimation method, and generate a power output scenario of the first wind farm, wherein the operating parameters include at least historical wind speed and output power;

[0010] Based on the greedy strategy and the K-means algorithm, the generated power output scenario of the first wind farm is reduced to obtain the power output scenario of the second wind farm;

[0011] Establishing a distribution network reconstruction model for a power output scenario of the second wind farm, wherein the distribution network reconstruction model includes an objective function and constraint conditions;

[0012] The distribution network reconstruction model is solved by combining an improved multi-objective Bayesian optimization algorithm.

[0013] As a preferred solution of the distribution network reconstruction method described in the present invention, wherein: the cumulative distribution function of each wind farm is established according to the operating parameters combined with the non-parametric estimation method, and the power output scenario of the first wind farm is generated, including: assuming that the number of wind farms is N, x1, x2, ..., x N , represents the output random variables of N wind farms at the same time, represents the random variables output by N wind farms at the same time, and generates the power output scenario of the first wind farm based on the inverse transformation method.

[0014] As a preferred solution of the distribution network reconstruction method described in the present invention, wherein: the power output scenario of the first wind farm generated is reduced based on the greedy strategy and the K-means algorithm to obtain the power output scenario of the second wind farm, including:

[0015] Set the number of initial clusters to 2, and select N initial points of the K-means algorithm based on the greedy strategy;

[0016] Execute the K-means clustering algorithm to assign samples in the data set to the nearest cluster center;

[0017] Determine whether the algorithm needs to be terminated;

[0018] If the termination condition is met, the clustering result is output;

[0019] Otherwise continue to increase the number of clusters and repeat the process.

[0020] As a preferred solution of the distribution network reconstruction method described in the present invention, the termination condition includes:

[0021] If the relative average error, relative standard deviation error, and relative Pearson coefficient parameter error between the generated scene and the restored scene are less than 10%, the termination condition is met, and the clustering result is output;

[0022] According to the obtained clustering results, the probability of each scenario reduction scheme is calculated according to the following formula:

[0023]

[0024] Among them, N g is the number of generated scenes, S Ng,i is the wind speed observation value of the i-th generated scene, μ ci is the wind speed observation value corresponding to the i-th scene, p i is the probability of the i-th scenario, and z represents a certain wind speed observation value.

[0025] As a preferred solution of the distribution network reconstruction method described in the present invention, wherein: the distribution network reconstruction model for the power output scenario of the second wind farm is established, and the distribution network reconstruction model includes an objective function and constraint conditions including:

[0026] The objective function is expressed as:

[0027]

[0028] Among them, Q(x,u,S) represents the objective function of the network topology optimization problem, V(x,u,S) represents the objective function of the wind power voltage deviation, L(x,u,S) represents the objective function of estimating the switch switching cost based on the total number of switching times, x is the switch state of the distribution network, u is the switch action signal, and S is the sample generated by the wind power scenario.

[0029] As a preferred solution of the distribution network reconstruction method described in the present invention, wherein: the distribution network reconstruction model for the power output scenario of the second wind farm is established, and the distribution network reconstruction model includes an objective function and constraints and also includes:

[0030] The objective function of the network topology optimization problem is expressed as:

[0031]

[0032] Among them, x is the switch state of the distribution network, u is the switch action signal, S is the sample generated by the wind power scenario, ρ i is the occurrence probability of wind power scenario i, Q i is the wind abandonment volume of scene i, and N is the total number of scenes;

[0033] The objective function of wind power voltage deviation is expressed as:

[0034]

[0035] Among them, K B is the number of distribution network nodes, V j is the voltage amplitude at node j, Vref is the reference stable voltage;

[0036] The objective function of estimating the switching cost based on the total number of switching times is:

[0037]

[0038] Where n(·) is the number of elements in the set and u0 is the initial switch set.

[0039] As a preferred solution of the distribution network reconstruction method of the present invention, the solving of the distribution network reconstruction model by combining the improved multi-objective Bayesian optimization algorithm includes:

[0040] A scenario reduction method is used to generate various wind power samples and the population is randomly initialized, where each particle represents a candidate solution;

[0041] For each circuit of the distribution network, a connecting switch is randomly selected, and its corresponding branch is disconnected. At the same time, the non-dominated archive η and the additional archive ζ are initialized;

[0042] In each iteration, each particle is used to calculate the wind power abandonment and voltage deviation, update η and ζ, and use the Bayesian network combined with the K2 algorithm to estimate the joint probability distribution of η;

[0043] After each iteration, a new population is drawn from χ;

[0044] If the number of iterations reaches the threshold, the algorithm terminates and the Pareto set ζ is obtained in the last iteration.

[0045] In a second aspect, the present invention provides a distribution network reconstruction system, comprising:

[0046] A data acquisition module, used to acquire operating parameters of different wind farms in the target distribution network, establish a cumulative distribution function of each wind farm based on the operating parameters in combination with a non-parametric estimation method, and generate a power output scenario of the first wind farm, wherein the operating parameters include at least historical wind speed and output power;

[0047] A scenario reduction module, used for reducing the generated power output scenario of the first wind farm based on a greedy strategy and a K-means algorithm to obtain a power output scenario of the second wind farm;

[0048] A model building module, used to build a distribution network reconstruction model for a power output scenario of the second wind farm, wherein the distribution network reconstruction model includes an objective function and constraint conditions;

[0049] The solution module is used to solve the distribution network reconstruction model in combination with an improved multi-objective Bayesian optimization algorithm.

[0050] In a third aspect, the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned method when executing the computer program.

[0051] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the method described above when executed by a processor.

[0052] Compared with the prior art, the invention has the following beneficial effects: the invention proposes a distribution network reconstruction method and system, obtains the operating parameters of different wind farms in the target distribution network, establishes the cumulative distribution function of each wind farm according to the operating parameters combined with the non-parametric estimation method, and generates the power output scenario of the first wind farm, wherein the operating parameters at least include the historical wind speed and the output power; based on the greedy strategy and the K-means algorithm, the generated power output scenario of the first wind farm is reduced to obtain the power output scenario of the second wind farm; establishes a distribution network reconstruction model for the power output scenario of the second wind farm, wherein the distribution network reconstruction model includes an objective function and constraints; and solves the distribution network reconstruction model in combination with the improved multi-objective Bayesian optimization algorithm. The method takes into account the problems of wind power absorption rate, voltage stability and system operation cost, and can ensure voltage stability while ensuring a high wind power absorption rate. The algorithm can still ensure a high convergence speed on the basis of a smaller population, and compared with other algorithms, the algorithm solution is closer to the optimal solution. In general, the algorithm improves the efficiency and accuracy of the algorithm. This method accurately characterizes the spatial correlation of the output power of multiple wind farms and improves the effectiveness of the generated scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0054] Figure 1 A method flow chart of a distribution network reconstruction method and system provided by an embodiment of the present invention;

[0055] Figure 2 A scene reduction flow chart based on a greedy strategy and a K-means algorithm for a distribution network reconstruction method and system provided by an embodiment of the present invention;

[0056] Figure 3A Pareto solution diagram of a distribution network reconstruction method and system provided by an embodiment of the present invention;

[0057] Figure 4 A flow chart of a multi-objective Bayesian optimization algorithm of a distribution network reconstruction method and system provided by an embodiment of the present invention;

[0058] Figure 5 A flowchart of an improved multi-objective Bayesian optimization algorithm for a distribution network reconstruction method and system provided by an embodiment of the present invention;

[0059] Figure 6 A schematic diagram of an improved IEEE-33 node system of a distribution network reconstruction method and system provided by an embodiment of the present invention;

[0060] Figure 7 An internal structural diagram of a computer device of a distribution network reconstruction method and system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0061] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0062] Example 1

[0063] Reference Figure 1-Figure 7 , is the first embodiment of the present invention, which provides a distribution network reconstruction method and system, including a distribution network reconstruction method and a distribution network reconstruction system, wherein a distribution network reconstruction method includes:

[0064] In the related technologies, some existing model studies have considered the voltage stability of the system and the absorption rate of wind power, but ignored the switching cost problem brought by the distribution network reconstruction to the system, resulting in excessively high reconstruction costs. In the existing algorithm research, due to the non-convexity of the reconstruction problem, it is difficult for the existing algorithms to find the optimal solution in the optimization problem. In the existing research, in terms of scenario generation, the spatial correlation of the power of multiple wind farms is almost not considered; at the same time, in terms of scenario reduction, the SBR method needs to continuously traverse the reserved scenarios, which has low computational efficiency and is mostly used for the reduction problem of the original scenario with a smaller scale.

[0065] The present application provides a method that can effectively solve the above-mentioned problems. Next, multiple embodiments will be combined to explain in detail how to implement the distribution network reconstruction method;

[0066] Figure 1 A method flow chart of a distribution network reconstruction method and system is shown, including:

[0067] S101, obtaining operating parameters of different wind farms in the target distribution network, establishing a cumulative distribution function of each wind farm based on the operating parameters in combination with a non-parametric estimation method, and generating a power output scenario of a first wind farm, wherein the operating parameters include at least a historical wind speed and an output power;

[0068] The operation parameters of different wind farms in the target distribution network are obtained, the cumulative distribution function of each wind farm is established according to the operation parameters combined with the non-parametric estimation method, and the power output scenario of the first wind farm is generated. Assuming that the number of wind farms is N, x1, x2, ..., x N , represents the random variables output by N wind farms at the same time, represents the random variables output by N wind farms at the same time, and generates the power output scenario of the first wind farm based on the inverse transformation method. Generation of multi-wind farm power generation scenarios based on multivariable Copula functions.

[0069] In the embodiment of the present application, the generation of a multi-wind farm power generation scenario based on a multivariate Copula function can be achieved by the following steps:

[0070] Step 1: Determine the cumulative distribution function of each wind farm power

[0071] There are two main methods to infer the probability distribution of a random variable from its observed value: parametric estimation and nonparametric estimation. Compared with parametric estimation, nonparametric estimation does not require prior knowledge of the sample distribution and only estimates the probability distribution density through the sample. Suppose x i (i=1,2,…,N) obeys continuous distribution if(xi), then its probability density function can be defined as:

[0072]

[0073] Where k is the number of samples, h is the fixed bandwidth, and K(*) is the kernel function. Common kernel functions include Gaussian kernel function, exponential kernel function, cosine kernel function, etc. Since the choice of kernel function has little effect on the estimation result, the Gaussian kernel function is selected as the kernel function:

[0074]

[0075] After selecting the Gaussian kernel function, the bandwidth of the Gaussian kernel function can be calculated by the following formula:

[0076]

[0077] After the Kennel function and bandwidth are given, the probability density function (CDF) of the random variable xi is known, and the cumulative probability distribution function of the random variable xi can be described as:

[0078]

[0079] Random variable x i The CDF value of the sample can be obtained by the following method:

[0080]

[0081] Step 2: Copula function parameter fitting and optimal Copula function selection

[0082] In step 1, for different multivariate Copula functions, the Copula function with the smallest distance to the empirical Copula function is considered to be the optimal Copula function:

[0083]

[0084] Among them, I[.] is a schematic function, if otherwise The Euclidean distance between the Copula function and the empirical Copula function can be described as:

[0085]

[0086] in, is the Copula function in (u 1,m ,…,u N,m ), C n (u 1,m ,…u N,m ) is the empirical Copula function in (u 1,m ,…,u N,m ) value.

[0087] Step 3: Scene generation method based on inverse transformation

[0088] After obtaining the optimal Copula function, the power output scenarios of multiple wind farms can be obtained based on the inverse transformation method. The implementation method of the inverse transformation method is as follows:

[0089] For a single wind farm i, [x isort,1 ,x isort,2 ,…,x isort,k ] is the historical output data sorted in ascending order, [u i,1 ,u i,2 ,…,u i,Ng ] is the CDF value generated by the optimal copula function, N gis the number of generated scenes. is the number of generated scenarios. Wind farm The output can be obtained as follows:

[0090]

[0091] For each wind farm, the inverse transformation method is used to implement a multi-wind farm scenario generation method that considers spatial correlation.

[0092] It should be noted that by considering the operating parameters of different wind farms, such as historical wind speed and output power, the cumulative distribution function of each wind farm is constructed in combination with non-parametric estimation methods. This method can more accurately simulate the actual operation of wind farms, especially when dealing with the complexity of multiple wind farms, and can capture the spatial correlation between different wind farms. By applying the inverse transformation method and multivariate Copula function, the generated power output scenario not only reflects the independent characteristics of each wind farm, but also takes into account the mutual influence between the outputs of each wind farm, thereby improving the realism of the scenario.

[0093] S102, reducing the generated power output scenario of the first wind farm based on a greedy strategy and a K-means algorithm to obtain a power output scenario of a second wind farm;

[0094] The power output scenario of the first wind farm is reduced based on the greedy strategy and the K-means algorithm, and the power output scenario of the second wind farm is obtained, including:

[0095] Set the number of initial clusters to 2, and select N initial points of the K-means algorithm based on the greedy strategy;

[0096] Execute the K-means clustering algorithm to assign samples in the data set to the nearest cluster center;

[0097] Determine whether the algorithm needs to be terminated;

[0098] If the termination condition is met, the clustering result is output;

[0099] Otherwise continue to increase the number of clusters and repeat the process.

[0100] Specifically, in this application, the termination conditions include:

[0101] If the relative average error, relative standard deviation error, and relative Pearson coefficient parameter error between the generated scene and the restored scene are less than 10%, the termination condition is met, and the clustering result is output;

[0102] In the embodiment of the present application, based on the obtained clustering results, the probability of each scenario reduction scheme is calculated according to the following formula:

[0103]

[0104] Among them, N g is the number of generated scenes, is the wind speed observation value of the i-th generated scene, is the wind speed observation value corresponding to the i-th scene, p i is the probability of the i-th scenario, and z represents a certain wind speed observation value.

[0105] It should be noted that the selection of the initial center point has an important impact on the results of the clustering algorithm. By selecting the appropriate initial center point, the number of restored scenes can be reduced as much as possible while ensuring the effect of scene restoration. The key to the greedy strategy is to determine the next optimal initial center point based on the optimal initial center point under the current determined number, and add it to the set of optimal initial center points. It can be predicted that as the number of clusters increases, the scene reduction results obtained by the improved K-means algorithm will be closer to the probability distribution characteristics of large-scale flow graphs of the greedy strategy and K-means algorithm, such as Figure 2 shown.

[0106] It should be noted that if the relative average error, relative standard deviation error and relative Pearson coefficient parameter error between the generated scenario and the restored scenario are less than ten percent, the termination condition is met. The constraints of the relative average error and relative standard deviation error are to ensure that the output of each wind farm after restoration conforms to its respective marginal distribution characteristics. The constraint of the relative Pearson correlation accuracy error is to ensure that the spatial correlation of the outputs of multiple wind farms can be accurately characterized. Specifically, if each relative error is less than ten percent, the reduced scheme is considered to be appropriate and conforms to the probability distribution of the generated scheme.

[0107] S103, establishing a distribution network reconstruction model for the power output scenario of the second wind farm, the distribution network reconstruction model including an objective function and constraint conditions;

[0108] In the embodiment of the present application, the distribution network reconstruction model can be expressed as:

[0109]

[0110] stdet(A)=±1

[0111] V i min ≤V i ≤V i max ,

[0112]

[0113]

[0114] Among them, G is the objective function that takes into account both voltage deviation and wind power consumption. is the vector representing the system state, N B is the number of nodes, V i and θ i are the voltage amplitude and phase angle respectively, u=[u1,u2,...,u n ] T is the variable vector for controlling the switch action of the distribution network, S is the wind power sample set obtained by the scenario reduction method. det(A)=±1 is the topological constraint, which physically means that all nodes are connected to the root node and there is no loop in the topology. V i min ≤V i ≤V i max and are the voltage and current safety constraints respectively, and is the power flow equation of the system. Since the power factor of the wind turbine generator used in the present invention is constant, a group of distributed generators can be regarded as a PQ node. In this way, the formula of injected power and required power is given by and It should be noted that this problem is a non-convex problem.

[0115] Among them, a distribution network reconstruction model for the power output scenario of the second wind farm is established, and the distribution network reconstruction model includes an objective function and constraint conditions including:

[0116] Specifically, the objective function is expressed as:

[0117]

[0118] Among them, Q(x,u,S) represents the objective function of the network topology optimization problem, V(x,u,S) represents the objective function of the wind power voltage deviation, L(x,u,S) represents the objective function of estimating the switch switching cost based on the total number of switching times, x is the switch state of the distribution network, u is the switch action signal, and S is the sample generated by the wind power scenario.

[0119] Furthermore, a distribution network reconstruction model for the power output scenario of the second wind farm is established. The distribution network reconstruction model includes an objective function and constraints, and also includes:

[0120] The distribution network reconstruction problem is essentially a dynamic grid planning problem of multi-objective optimization. Both the topological level optimization and the time level optimization can be transformed into a mathematical problem of nonlinear integer optimization. The optimization variables are the distribution network action switch combination set and the reconstruction cycle segment number set.

[0121] In the problem of wind power access distribution network reconstruction, in order to improve the distribution network's absorption rate of wind energy, it is necessary to find a topology with smaller losses. Considering the uncertainty of wind power, the objective function of the network topology optimization problem is expressed as:

[0122]

[0123] Among them, x is the switch state of the distribution network, u is the switch action signal, S is the sample generated by the wind power scenario, ρ i is the occurrence probability of wind power scenario i, Q i is the wind abandonment volume of scene i, and N is the total number of scenes;

[0124] Bus voltage within an appropriate range is very important for the safe operation of the distribution network. Overvoltage and undervoltage will affect the stable operation of the power grid. When other conditions remain unchanged, the smaller the voltage deviation, the more conducive to the stable operation of the power grid. The objective function of wind power voltage deviation is expressed as:

[0125]

[0126] Among them, K B is the number of distribution network nodes, V j is the voltage amplitude at node j, V ref is the reference stable voltage;

[0127] In addition to wind abandonment and voltage deviation, frequent switching has an adverse effect on the economic operation of the distribution network. Therefore, the present invention uses the total number of switching times to estimate the switching cost. The objective function of the total number of switching times to estimate the switching cost is:

[0128]

[0129] Where n(·) is the number of elements in the set and u0 is the initial switch set.

[0130] It should be noted that this problem comprehensively considers the amount of abandoned air, voltage bias and operating costs generated by the switch, and the present invention takes this as the objective function of the research.

[0131] S104, solving the distribution network reconstruction model by combining an improved multi-objective Bayesian optimization algorithm.

[0132] Among them, the solution of the distribution network reconstruction model by combining the improved multi-objective Bayesian optimization algorithm includes:

[0133] A scenario reduction method is used to generate various wind power samples and the population is randomly initialized, where each particle represents a candidate solution;

[0134] For each circuit of the distribution network, a connecting switch is randomly selected, and its corresponding branch is disconnected. At the same time, the non-dominated archive η and the additional archive ζ are initialized;

[0135] In each iteration, each particle is used to calculate the wind power abandonment and voltage deviation, update η and ζ, and use the Bayesian network combined with the K2 algorithm to estimate the joint probability distribution of η;

[0136] After each iteration, a new population is drawn from χ;

[0137] If the number of iterations reaches the threshold, the algorithm terminates and the Pareto set ζ is obtained in the last iteration.

[0138] In the embodiment of the present application, a Pareto solution of a multi-objective optimization problem is established, wherein a multi-objective optimization problem can generally be expressed as:

[0139]

[0140] sth(u)≤0

[0141] g(u)=0

[0142] in, The feasible solutions are target value.

[0143] In existing research, multi-objective evolutionary computing methods are often used to solve such problems. In addition, weighted methods are often used to transform multi-objective optimization problems into single-objective optimization problems to solve such problems. However, the above methods are not applicable to the distribution network reconstruction problem proposed in this invention. There are two main reasons. On the one hand, the weights in the weighted method are difficult to select; on the other hand, the model of this invention is non-convex and it is difficult to find an analytical solution.

[0144] In the embodiment of the present application, a Bayesian algorithm based on Pareto sets and non-dominated solutions is used to solve such problems. To clearly describe this algorithm, the following concepts are added here:

[0145] Dominant solution: For any k, And there exists m such that Then u j by u i Domination, that is, u i ≤u j In addition, if u i is not dominated by other solutions, then u i is a non-dominated solution. For ease of understanding, here is an example, Figure 3 As shown, the solution in the dashed box is the dominant solution, and the solution outside the dashed box is the non-dominated solution;

[0146] Pareto set: The Pareto set consists of non-dominated solutions, that is, μ = {u 1 ,u 2 ,...,u N};

[0147] Pareto frontier: The multi-objective function value of the Pareto set is composed of the Pareto frontier, that is, ρ = {G 1 ,G 2 ,...,G N}.

[0148] In the embodiment of the present application, the iterative process of the multi-objective Bayesian optimization algorithm is as follows: Figure 4 As shown, the algorithm uses a population That is, iteratively search for non-dominated solutions from N0 candidate solutions. Then, all non-dominated solutions are saved in a non-dominated archive, represented by η. After that, the algorithm uses the Bayesian network χ={V,E} to estimate the joint probability distribution of the obtained non-dominated solutions. Among them, a vertex in V represents one dimension in u, and an edge in E represents the related two dimensions. It should be noted that the K2 algorithm in the figure is used to determine whether there is a correlation between two vertices, that is, whether there is an edge in E. The metric formula of the algorithm is:

[0149]

[0150] Among them, ν i and π i denote vertices i and ν in V respectively i The parent vertex of ν i The number of possible values ​​of Represents π i have possible values. Among them, n(·) represents the number of corresponding samples in η. If the existence of an edge brings greater benefits to the function, then this edge is retained. Based on this, the joint probability distribution of the Bayesian network can be obtained as follows:

[0151]

[0152] Where p represents vertex π i is the vertex ν i The probability of the parent vertex. After that, a new population can be generated by stratified sampling of the obtained Bayesian network. Specifically, the values ​​of the randomly generated vertices without parents are first processed. Then, the values ​​of the child vertices are obtained according to the conditional probability. When the algorithm visits all vertices, a new sample is obtained.

[0153] In the embodiment of the present application, in order to better approach the true distribution, it is necessary to obtain a sufficiently large sample from η. For this reason, the multi-objective Bayesian optimization algorithm uses a population with a larger N0. However, this method also generates huge computational costs. Therefore, obtaining enough samples in a smaller population becomes an important issue.

[0154] The multi-objective Bayesian optimization algorithm obtains the Bayesian fitting non-dominated archive η shown in the formula. However, η is updated through iteration, in which repeated solutions and new dominant solutions are removed from η. It should be pointed out that these removed solutions do not contribute much to the establishment of the Bayesian network. Therefore, in order to make full use of the information in these solutions, it is necessary to leave these solutions to generate enough samples.

[0155] In order to make full use of the information in the original algorithm solution, the present invention designs an additional archive ζ to save these solutions and improve the original algorithm. Of course, the capacity of ζ is limited, expressed as M ζ After each iteration to obtain η, the algorithm will count the number of solutions N in the archive ζ ζ Evaluate and update according to the following rules:

[0156] If N ζ <M ζ , then ζ←ζ∪η;

[0157] If N ζ ≥M ζ , then remove N from ζ ζ -M ζ A dominant solution, then, ζ←ζ∪η.

[0158] Then, the improved multi-objective Bayesian optimization algorithm uses the Bayesian network to estimate the joint probability distribution of the solution in ζ, and its formula is:

[0159]

[0160] In the embodiment of the present application, the process of improving the multi-objective Bayesian optimization algorithm is as follows: Figure 5 As shown in the figure, similar to the original algorithm, the proposed algorithm uses μ={u1,u2,...,u N}Search for non-dominated solutions, where N<<N0. The non-dominated solutions selected from the population are saved in the non-dominated archive η, and then an additional archive ζ is added to save a copy of η. At the same time, the number of solutions in ζ is updated by the above two rules. After that, a Bayesian network is used to generate a joint probability distribution of ζ, and a new population is extracted from the obtained Bayesian network. Repeat the above steps until the maximum number of iterations is reached. Finally, the non-dominated solutions in η are used to construct an approximate Pareto set. By Figure 5 It can be concluded that only in M ζAfter reaching the maximum value, the solution in ζ will be removed. Therefore, some feasible solutions may have multiple copies in ζ. In this way, enough samples can be obtained to build a Bayesian network at a small computational cost.

[0161] In the embodiment of the present application, in order to apply the improved multi-objective Bayesian optimization algorithm to the distribution network reconstruction model, it is necessary to calculate the objective function. It should be noted that all objective functions are calculated under the minimum curtailment state described in detail in Algorithm 1. After the minimum wind abandonment is obtained, the corresponding objectives can be obtained by calculating different objective functions respectively.

[0162] The specific process of solving the distribution network reconstruction problem is shown in Algorithm 2. Initially, the present invention uses a scenario reduction method to generate various wind power samples, and then randomly initializes the population, in which each particle represents a candidate solution. Specifically, for each loop of the distribution network, a connecting switch is randomly selected, and its corresponding branch is disconnected. At the same time, the non-dominated archive η and the additional archive ζ are initialized. In each iteration, each particle is used to calculate the wind power abandonment and voltage deviation. Then, η and ζ are updated, and the joint probability distribution of η is estimated using the Bayesian network combined with the K2 algorithm. After each iteration, a new population is extracted from χ. If the number of iterations reaches the threshold, the algorithm terminates, and the Pareto set ζ is obtained in the last iteration. It is worth emphasizing that the present invention uses the TOPSIS method to select the final solution from the Pareto set.

[0163]

[0164]

[0165]

[0166]

[0167]

[0168] In an optional embodiment, after the last iteration, a Pareto optimal solution set is obtained from the non-dominated archive. These solution sets contain multiple distribution network reconstruction schemes that balance the best among the abandoned wind volume, voltage deviation and switch switching cost. According to actual needs and system constraints, a most suitable reconstruction scheme is selected from the Pareto optimal solution set for implementation. This scheme will guide the adjustment of the switch state of the distribution network and optimize the network topology to reduce the abandoned wind volume, improve voltage stability, and control the switch operation cost.

[0169] In summary, the present invention proposes a distribution network reconstruction method, which obtains the operating parameters of different wind farms in the target distribution network, establishes the cumulative distribution function of each wind farm according to the operating parameters combined with the non-parametric estimation method, and generates a power output scenario of the first wind farm, wherein the operating parameters at least include historical wind speed and output power; based on the greedy strategy and K-means algorithm, the generated power output scenario of the first wind farm is reduced to obtain the power output scenario of the second wind farm; a distribution network reconstruction model for the power output scenario of the second wind farm is established, wherein the distribution network reconstruction model includes an objective function and constraints; and the distribution network reconstruction model is solved in combination with an improved multi-objective Bayesian optimization algorithm. This method takes into account the problems of wind power absorption rate, voltage stability and system operating cost, and can ensure voltage stability while ensuring a high wind power absorption rate. The algorithm

[0170] This embodiment also provides a distribution network reconstruction system, including:

[0171] A data acquisition module is used to obtain the operating parameters of different wind farms in the target distribution network, establish a cumulative distribution function of each wind farm based on the operating parameters in combination with a non-parametric estimation method, and generate a power output scenario of the first wind farm, wherein the operating parameters include at least historical wind speed and output power;

[0172] A scenario reduction module, used for reducing the generated power output scenario of the first wind farm based on a greedy strategy and a K-means algorithm to obtain a power output scenario of the second wind farm;

[0173] A model building module, used to build a distribution network reconstruction model for the power output scenario of the second wind farm, the distribution network reconstruction model including an objective function and constraint conditions;

[0174] The solution module is used to solve the distribution network reconstruction model in combination with the improved multi-objective Bayesian optimization algorithm.

[0175] The above-mentioned unit modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above-mentioned modules.

[0176] This embodiment also provides a computer device, which may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 7As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a distribution network reconstruction method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0177] This embodiment further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0178] Obtaining operating parameters of different wind farms in the target distribution network, establishing a cumulative distribution function of each wind farm based on the operating parameters in combination with a non-parametric estimation method, and generating a power output scenario of the first wind farm, wherein the operating parameters include at least historical wind speed and output power;

[0179] Based on the greedy strategy and the K-means algorithm, the generated power output scenario of the first wind farm is reduced to obtain the power output scenario of the second wind farm;

[0180] Establishing a distribution network reconstruction model for a power output scenario of the second wind farm, the distribution network reconstruction model including an objective function and constraint conditions;

[0181] The distribution network reconstruction model is solved by combining the improved multi-objective Bayesian optimization algorithm.

[0182] Example 2

[0183] Reference Figure 2-Figure 6 , which is an embodiment of the present invention, provides a distribution network reconstruction method and system. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0184] The present invention verifies the model using an improved IEEE-33 node system, which consists of 33 nodes and 37 branches. It is assumed that each wind turbine has the same dynamic performance, that is, a doubly fed induction generator with a constant power factor. The rated power is 17.5KW, the cut-in speed is 4m / s, the rated speed is 12.5m / s, the cut-off speed is 20m / s, and the number of wind power samples after the scenario reduction is set to 30.

[0185] In order to prove the advantages of the proposed algorithm, three algorithms are introduced and applied to the distribution network reconstruction problem for comparison. These algorithms include multi-objective particle swarm optimization (MOPSO), multi-objective evolutionary algorithm based on decomposition (MOEA / D), and strength Pareto evolutionary algorithm (SPEA). Specifically, the verification process uses 30 independent runs to compare the performance of the four algorithms. Among them, the ε of the Newton-Raphson algorithm is set to p For 10 -6 In addition, the population size N is set to 30 and the maximum number of iterations M is set to 30. For the improved multi-objective Bayesian optimization algorithm, in order to accommodate enough historical non-dominated solutions, the capacity of the additional archive M ζ Set to 1000.

[0186] The present invention uses four indicators to measure the evolutionary algorithm, namely, hyper-volume (HV), mean distance (MD), and spacing (SP) index. More precisely, HV is the volume covered by non-dominated solutions when a reference point is given, and is an indicator for evaluating the convergence and diversity of the obtained Pareto frontier. MD represents the average Euclidean distance from the Pareto frontier to the reference point, representing the convergence of the algorithm. SP is used to estimate diversity, and the search efficiency index is quantified by the number of solutions of the Pareto set (NSPS). The HV, MD, and SP indices are calculated by the formulas respectively.

[0187]

[0188] Where r0 is the reference point, Nη is the number of solutions in η, and Vol(·) represents the volume between two points. Therefore, larger HV and MD values ​​indicate better Pareto front convergence, while smaller SP values ​​indicate better Pareto front distribution. Therefore, a good solution selection is a compromise between a fast convergence rate and a well-distributed Pareto front.

[0189] In addition, mixed integer programming (MIP) is a common method for dealing with distribution network reconstruction models, especially with the advantage of ensuring local optimal solutions. Therefore, MIP is also applicable to the proposed distribution network reconstruction model. In this problem, the algorithm uses the weighted sum method to merge multiple objectives, and the weight of each objective function is equal. Specifically, the model is implemented in GAMS and the MINLP solver SBB is used to solve the model.

[0190] Test system such as Figure 6 The improved IEEE-33 node system is shown in Figure 1. The standard deviation of wind speed prediction error is set to 8% of the predicted wind speed.

[0191] In order to examine the performance of the distribution network reconstruction model proposed in the present invention, two objective functions of the initial state and the distribution network reconstruction are listed in Table 1. It can be seen that the wind power abandonment in the initial state is 147kW, which is much larger than the wind power abandonment in the distribution network reconstruction model. At the same time, the voltage deviation is reduced from 0.0384 to 0.0267. Therefore, the distribution network reconstruction model improves the absorption rate of wind power while maintaining voltage stability, which is ideal in practical applications.

[0192] Table 1 Objective function of the scheduling scheme calculated using the model proposed by the present invention and MIP

[0193]

[0194] Table 2 Comparison of three Pareto frontier distributions of four algorithms

[0195]

[0196] Table 1 Comparison of five algorithms including HV, MD and NSPS

[0197]

[0198] To compare the proposed algorithm with the other three algorithms, Table 2 lists the average distances between the Pareto set and the panel and the SP index, with the minimum value marked in bold. It is worth noting that the proposed algorithm is almost the closest to all xz, xy, and yz planes (only slightly higher than the distance from MOEA / D to the yz plane), which indicates that it has a smaller wind power abandonment and a smaller voltage deviation. At the same time, the lower the SP, the more evenly distributed the solutions are. Therefore, the proposed algorithm is superior to other algorithms in finding evenly distributed solutions.

[0199] In order to quantify the Pareto frontier of the algorithm proposed in the present invention, the reference point r0 is set to (1.1, 0.10, 10), which can also be replaced by other points. The comparison of the three indicators is shown in Table 3. More precisely, in HV, the value of the algorithm proposed in the present invention is the largest, indicating that it converges to the Pareto frontier the fastest. At the same time, the high MD of the algorithm proposed in the present invention means a more convergent Pareto optimal solution. The algorithm proposed in the present invention obtains 39 solutions, which is close to MOPSO and larger than other methods. This verifies the search advantage of the algorithm proposed in the present invention.

[0200] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0201] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0202] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0203] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0204] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0205] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0206] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A distribution network reconstruction method, characterized in that: include: Acquire operating parameters of different wind farms in the target distribution network, establish a cumulative distribution function of each wind farm based on the operating parameters in combination with a non-parametric estimation method, and generate a power output scenario of the first wind farm, wherein the operating parameters include at least historical wind speed and output power; Based on the greedy strategy and the K-means algorithm, the generated power output scenario of the first wind farm is reduced to obtain the power output scenario of the second wind farm; Establishing a distribution network reconstruction model for a power output scenario of the second wind farm, wherein the distribution network reconstruction model includes an objective function and constraint conditions; The distribution network reconstruction model is solved by combining an improved multi-objective Bayesian optimization algorithm.

2. The distribution network reconstruction method according to claim 1, characterized in that: The step of obtaining the operating parameters of different wind farms in the target distribution network, establishing the cumulative distribution function of each wind farm according to the operating parameters in combination with the non-parametric estimation method, and generating the power output scenario of the first wind farm includes: assuming that the number of wind farms is N, x1, x2, ..., x N , represents the output random variables of N wind farms at the same time, represents the random variables output by N wind farms at the same time, and generates the power output scenario of the first wind farm based on the inverse transformation method.

3. The distribution network reconstruction method according to claim 2, characterized in that: The power output scenario of the first wind farm generated is reduced based on the greedy strategy and the K-means algorithm to obtain the power output scenario of the second wind farm, which includes: Set the number of initial clusters to 2, and select N initial points of the K-means algorithm based on the greedy strategy; Execute the K-means clustering algorithm to assign samples in the data set to the nearest cluster center; Determine whether the algorithm needs to be terminated; If the termination condition is met, the clustering result is output; Otherwise continue to increase the number of clusters and repeat the process.

4. The distribution network reconstruction method according to claim 3, characterized in that: The termination conditions include: If the relative average error, relative standard deviation error, and relative Pearson coefficient parameter error between the generated scene and the restored scene are less than 10%, the termination condition is met, and the clustering result is output; According to the obtained clustering results, the probability of each scenario reduction scheme is calculated according to the following formula: Among them, N g is the number of generated scenes, is the wind speed observation value of the i-th generated scene, is the wind speed observation value corresponding to the i-th scene, p i is the probability of the i-th scenario, and z represents a certain wind speed observation value.

5. The method for reconfiguring a distribution network according to claim 4, characterized in that: The distribution network reconstruction model for the power output scenario of the second wind farm is established, and the distribution network reconstruction model includes an objective function and constraint conditions including: The objective function is expressed as: Among them, Q(x,u,S) represents the objective function of the network topology optimization problem, V(x,u,S) represents the objective function of the wind power voltage deviation, L(x,u,S) represents the objective function of estimating the switch switching cost based on the total number of switching times, x is the switch state of the distribution network, u is the switch action signal, and S is the sample generated by the wind power scenario.

6. The distribution network reconstruction method according to claim 5, characterized in that: The distribution network reconstruction model for the power output scenario of the second wind farm is established, wherein the distribution network reconstruction model includes an objective function and constraint conditions and further includes: The objective function of the network topology optimization problem is expressed as: Among them, x is the switch state of the distribution network, u is the switch action signal, S is the sample generated by the wind power scenario, ρ i is the occurrence probability of wind power scenario i, Q i is the abandoned air volume in scene i, and N is the total number of scenes; The objective function of wind power voltage deviation is expressed as: Among them, K B is the number of distribution network nodes, V j is the voltage amplitude at node j, V ref is the reference stable voltage; The objective function of estimating the switching cost based on the total number of switching times is: Where n(·) is the number of elements in the set and u0 is the initial switch set.

7. The method for reconfiguring a distribution network according to claim 6, characterized in that: The solving of the distribution network reconstruction model by combining the improved multi-objective Bayesian optimization algorithm includes: A scenario reduction method is used to generate various wind power samples and the population is randomly initialized, where each particle represents a candidate solution; For each circuit of the distribution network, a connecting switch is randomly selected, and its corresponding branch is disconnected. At the same time, the non-dominated archive η and the additional archive ζ are initialized; In each iteration, each particle is used to calculate the wind power abandonment and voltage deviation, update η and ζ, and use the Bayesian network combined with the K2 algorithm to estimate the joint probability distribution of η; After each iteration, a new population is drawn from χ; If the number of iterations reaches the threshold, the algorithm terminates and the Pareto set ζ is obtained in the last iteration.

8. A distribution network reconstruction system, characterized in that: include: A data acquisition module, used to acquire operating parameters of different wind farms in the target distribution network, establish a cumulative distribution function of each wind farm based on the operating parameters in combination with a non-parametric estimation method, and generate a power output scenario of the first wind farm, wherein the operating parameters include at least historical wind speed and output power; A scenario reduction module, used for reducing the generated power output scenario of the first wind farm based on a greedy strategy and a K-means algorithm to obtain a power output scenario of the second wind farm; A model building module, used to build a distribution network reconstruction model for a power output scenario of the second wind farm, wherein the distribution network reconstruction model includes an objective function and constraint conditions; The solution module is used to solve the distribution network reconstruction model in combination with an improved multi-objective Bayesian optimization algorithm.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images