Power distribution network topological structure analysis method
By preprocessing the distribution network node data and introducing ant colony algorithm to optimize the universal gravitational search model, the problem of insufficient computing complexity and generalization capabilities in the topological structure analysis of the distribution network is solved, and more efficient topological structure analysis and Stronger generalization ability.
Patent Information
- Application Number
- CN202411626618.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-14
- Publication Date
- 2025-02-18
AI Technical Summary
Among the existing distribution network topological structure analysis methods, the universal gravitational search algorithm has a high computational complexity and is prone to fall into local optimal solutions, resulting in poor generalization capabilities and cannot effectively solve the computational complexity and generalization capabilities in distribution network topological structure analysis. A problem of insufficient.
By preprocessing the distribution network node data, and introducing an ant colony algorithm to optimize the universal gravitational search model, using the two types of methods of fitness function and particle division, the ant colony algorithm is used to optimize the speed of the first type of particles, and improve the convergence of the model Speed and generalization ability.
It effectively reduces the computational complexity of the universal gravitational search model, avoids the problem of falling into local optimal solutions, improves the generalization ability of the model, and makes it suitable for topological analysis applications of various distribution networks.
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of power grid topology, and in particular to a method for analyzing a distribution network topology structure. Background Art
[0002] The topological analysis of the distribution network is the process of analyzing the topological logic of power supply nodes, power supply contacts and switch nodes as nodes in the topological diagram, which is used to estimate the working status of the distribution network and optimize the structure of the distribution network. The topological analysis method of the distribution network has gradually developed from the initial manual measurement test and the calculation of the topological structure through Kirchhoff's law to computer simulation and artificial intelligence analysis. At present, various machine learning models are usually used to analyze the topological structure of the distribution network. Although the analysis accuracy and efficiency are significantly improved, a large amount of high-quality data is required for model training, and the generalization ability is poor.
[0003] The gravitational search algorithm is a heuristic algorithm developed in recent years. The particles in the algorithm are regarded as a group of objects moving in space. They interact with each other through gravitation. The object with the largest mass occupies the optimal position, thus obtaining the optimal solution. The gravitational search algorithm has the advantages of few parameters, easy adjustment, and strong generalization ability. However, in the topological analysis process of the distribution network, the computational complexity is high and it is easy to fall into the local optimal solution. Therefore, it is not effective to directly apply it to the topological structure analysis of the distribution network. Summary of the invention
[0004] The technical problem to be solved by the present invention is to provide a distribution network topology structure analysis method, which can solve the shortcomings of the existing technology and achieve the purpose of improving the generalization ability of the topology analysis model by introducing an optimized universal gravitation search algorithm.
[0005] In order to solve the above technical problems, the technical solutions adopted by the present invention are as follows.
[0006] A distribution network topology analysis method comprises the following steps: A. Read the geographical location information of the distribution network node and the topological information of the connection with the adjacent distribution network nodes; B. Preprocessing the geographical location information and topology information of the distribution network nodes; C. Establish a gravitational search model, input the preprocessed distribution network node information into the gravitational search model for optimization, and obtain the optimized distribution network topology structure.
[0007] Preferably, in step B, preprocessing the geographical location information and topology information of the distribution network nodes includes the following steps: The geographic location information and topology information are classified to obtain several sub-category information, all the sub-category information are normalized to obtain the normalized distribution network node information, and then the normalized distribution network node information is decomposed in multiple dimensions.
[0008] Preferably, the geographical location information is classified in two dimensions according to the installation location distance of the distribution network nodes and the power consumption characteristics of the distribution network nodes, and the topology information is classified according to different parent nodes in the topology information.
[0009] Preferably, the distribution network nodes whose geographic location information classification and topology information classification belong to the same category are taken as reference nodes, the common features of the reference nodes are extracted, the mean and standard deviation of the common features are calculated, and all subclass information is normalized using the Z score.
[0010] Preferably, the normalized distribution network node information is multi-dimensionally expanded, and the expansion dimension is determined by the dimension of the original information of the distribution network node; in each expanded dimension, the connection direction between adjacent data points is used as the expansion direction, and the data points are copied along the expansion directions on both sides, that is, the weighted data point data is copied at the coordinate point in the expansion direction, and the weighted weight of the data is inversely proportional to the distance between the target coordinates of the copy and the data point coordinates, and then the data in each dimension is smoothed; in each dimension, a sliding window is used to traverse the smoothed data to mark the data stationary points.
[0011] Preferably, in step C, the preprocessed current distribution network node information is input as input data into the gravitational search model for optimization iteration, and the fitness function is f=k1f1+ k2f2+ k3f3+λp, wherein f1 is the power transmission loss function of the power grid, f2 is the power grid frequency volatility function, f3 is the power grid voltage volatility function, k1-k3 are the weights of the corresponding function components, p is the penalty term, λ is the penalty factor, and after each optimization iteration, the particle speed is corrected by the ant colony algorithm.
[0012] Preferably, the data stationary points are taken as first-class particles, and the remaining data are taken as second-class particles. The inertial mass of the first-class particles is greater than that of the second-class particles. The speed of the first-class particles is corrected by the ant colony algorithm. In order to simplify the algorithm calculation process, the inertial mass of the particles is equated with the gravitational mass of the particles.
[0013] Preferably, the minimum sum of the fitness f of all first-class particles is used as the optimization target of the ant colony algorithm, the possible speed size and speed direction of the first-class particles are used as the optimization range of the ant colony, and the speed size and speed direction of each first particle are used as the two parameters of the ant's optional optimization path. Each ant releases pheromones when passing through the path it chooses. After each ant passes through the path it chooses, the pheromone concentration of the path is evaporated. When all ants complete a path selection, the pheromone concentration on all paths is updated. The concentration of pheromones on the path is proportional to the probability that the ant chooses it as the optimization path. When the maximum number of iterations is reached or the optimal path is stable, the current optimal path is output as the corrected speed size and speed direction of the first particle.
[0014] The beneficial effect brought about by adopting the above technical solution is that: the present invention solves the problem of high computational complexity and easy falling into local optimal solution of the gravitational search model by preprocessing the power grid node data and introducing the ant colony algorithm into the gravitational search model, making it widely applicable to topological analysis applications of various distribution networks and improving the generalization ability of existing topological analysis models. DETAILED DESCRIPTION
[0015] In the description of the following embodiments, specific details such as specific system structures, technologies, etc. are provided for illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.
[0016] Due to the large number of topological nodes in the distribution network and the frequent changes in operating parameters, traditional optimization models such as genetic algorithms require a large amount of training data when used. Due to poor generalization ability, if the topology structure changes, retraining is required, which obviously limits the use of optimization models. In this regard, the gravitational search algorithm has a relative advantage due to the characteristics of its optimization logic. However, as mentioned above, the optimization calculation amount of the gravitational search algorithm is large, and there will be obvious response lag for the needs of distribution network topology analysis. To address this problem, we optimized the use process of the gravitational search algorithm, which includes the following steps: A. Read the geographical location information of the distribution network node and the topological information of the connection with the adjacent distribution network nodes.
[0017] B. Preprocess the geographical location information and topology information of the distribution network nodes; specifically: The geographical location information is classified in two dimensions according to the installation location distance of the distribution network nodes and the power consumption characteristics of the distribution network nodes, and the topology information is classified according to different parent nodes in the topology information.
[0018] The distribution network nodes whose geographic location information classification and topology information classification belong to the same category are taken as reference nodes. The common features of the reference nodes are extracted, the mean and standard deviation of the common features are calculated, and all subclass information is normalized using the Z score.
[0019] The normalized distribution network node information is multi-dimensionally expanded, and the expansion dimension is determined by the dimension of the original information of the distribution network node; in each expanded dimension, the connection direction between adjacent data points is used as the expansion direction, and the data points are copied along the expansion directions on both sides, that is, the weighted data point data is copied at the coordinate point in the expansion direction, and the weighted weight of the data is inversely proportional to the distance between the target coordinates of the copy and the coordinates of the data point, and then the data in each dimension is smoothed; in each dimension, a sliding window is used to traverse the smoothed data and mark the data stationary points.
[0020] The purpose of data preprocessing is to decouple different types of data in terms of data dimensions, so as to reduce the amount of data interaction calculations in different dimensions during the gravitational search process. By using common features for normalization, the feature information in the original data can be effectively preserved. After multi-dimensional expansion of the data, data can be decoupled and stable points of decoupled data of different dimensions can be extracted. The determination of stable points facilitates the introduction of the ant colony algorithm.
[0021] C. Establish a gravitational search model, and input the preprocessed current distribution network node information as input data into the gravitational search model for optimization iteration. The fitness function is f=k1f1+ k2f2+ k3f3+λp, where f1 is the power transmission loss function of the power grid, f2 is the power grid frequency volatility function, f3 is the power grid voltage volatility function, k1-k3 are the weights of the corresponding function components, p is the penalty term, λ is the penalty factor, and the data stationary point is regarded as the first type of particles, and the rest of the data is regarded as the second type of particles. The inertial mass of the first type of particles is greater than that of the second type of particles. In order to simplify the algorithm calculation process, the inertial mass of the particles is equal to the gravitational mass of the particles. After each optimization iteration, the velocity of the first type of particles is corrected by the ant colony algorithm. The minimum sum of the fitness f of all first-class particles is taken as the optimization goal of the ant colony algorithm, the possible speed size and speed direction of the first-class particles are taken as the optimization range of the ant colony, and the speed size and speed direction of each first particle are used as the optional optimization path of the ant. Each ant releases pheromones when passing through the path it chooses. After each ant passes through the path it chooses, the pheromone concentration of the path is evaporated. When all ants complete a path selection, the pheromone concentration on all paths is updated. The concentration of pheromones on the path is proportional to the probability that the ant chooses it as the optimization path. When the maximum number of iterations is reached or the optimal path is stable, the current optimal path is output as the corrected speed size and speed direction of the first particle.
[0022] We divide the particles in the gravitational search model into two categories, so that they form a structure similar to a multi-star system. The first type of particles is equivalent to the stars in the system. They have a large mass and have a significant gravitational effect on the movement of the surrounding small mass second type particles. Then the ant colony algorithm is introduced to optimize the movement trend of the first type of particles, and the optimization of the first type of particles is used to achieve a common influence on the second type of particles, thereby accelerating the convergence speed of the gravitational search model and avoiding falling into a local optimal solution. At the same time, because only the first type of particles need to be optimized, the amount of calculation of the ant colony algorithm optimization process introduced can be effectively controlled to avoid excessive calculations. After our optimization, the gravitational search model in this application not only has a faster effect speed, but also because the optimal solution range of the position of the first type of particles is relatively concentrated, the generalization ability of the model is greatly improved, and it is suitable for the topological analysis process of various types of distribution networks.
[0023] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.
[0024] The present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of each of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electric carrier signals and telecommunication signals.
[0025] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.
Claims
1. A distribution network topology analysis method, characterized in that The following steps are involved: A. Read the geographical location information of the distribution network node and the topological information of the connection with the adjacent distribution network nodes; B. Preprocessing the geographical location information and topology information of the distribution network nodes; C. Establish a gravitational search model, input the preprocessed distribution network node information into the gravitational search model for optimization, and obtain the optimized distribution network topology structure.
2. The distribution network topology analysis method according to claim 1, characterized in that: In step B, preprocessing the geographical location information and topology information of the distribution network nodes includes the following steps: The geographic location information and topology information are classified to obtain several sub-category information, all the sub-category information are normalized to obtain the normalized distribution network node information, and then the normalized distribution network node information is decomposed in multiple dimensions.
3. The method for analyzing the topological structure of a distribution network according to claim 2, characterized in that: The geographical location information is classified in two dimensions according to the installation location distance of the distribution network nodes and the power consumption characteristics of the distribution network nodes, and the topology information is classified according to different parent nodes in the topology information.
4. The method for analyzing the topological structure of a distribution network according to claim 3, characterized in that: The distribution network nodes whose geographic location information classification and topology information classification belong to the same category are taken as reference nodes. The common features of the reference nodes are extracted, the mean and standard deviation of the common features are calculated, and all subclass information is normalized using the Z score.
5. The method for analyzing the topological structure of a distribution network according to claim 4, characterized in that: The normalized distribution network node information is multi-dimensionally expanded, and the expansion dimension is determined by the dimension of the original information of the distribution network node; in each expanded dimension, the connection direction between adjacent data points is used as the expansion direction, and the data points are copied along the expansion directions on both sides, that is, the weighted data point data is copied at the coordinate point in the expansion direction, and the weighted weight of the data is inversely proportional to the distance between the target coordinates of the copy and the coordinates of the data point, and then the data in each dimension is smoothed; in each dimension, a sliding window is used to traverse the smoothed data and mark the data stationary points.
6. The method for analyzing the topological structure of a distribution network according to claim 5, characterized in that: In step C, the preprocessed current distribution network node information is input as input data into the gravitational search model for optimization iteration. The fitness function is f=k1f1+ k2f2+ k3f3+λp, where f1 is the power transmission loss function of the power grid, f2 is the power grid frequency volatility function, f3 is the power grid voltage volatility function, k1-k3 are the weights of the corresponding function components, p is the penalty term, and λ is the penalty factor. After each optimization iteration, the particle speed is corrected by the ant colony algorithm.
7. The method for analyzing the topological structure of a distribution network according to claim 6, characterized in that: The data station point is regarded as the first type of particles, and the rest of the data is regarded as the second type of particles. The inertial mass of the first type of particles is greater than that of the second type of particles. The speed of the first type of particles is corrected by the ant colony algorithm. In order to simplify the algorithm calculation process, the inertial mass of the particles is equated with the gravitational mass of the particles.
8. The method for analyzing the topological structure of a distribution network according to claim 7, characterized in that: The minimum sum of the fitness f of all first-class particles is taken as the optimization goal of the ant colony algorithm, the possible speed size and speed direction of the first-class particles are taken as the optimization range of the ant colony, and the speed size and speed direction of each first particle are used as the optional optimization path of the ant. Each ant releases pheromones when passing through the path it chooses. After each ant passes through the path it chooses, the pheromone concentration of the path is evaporated. When all ants complete a path selection, the pheromone concentration on all paths is updated. The concentration of pheromones on the path is proportional to the probability that the ant chooses it as the optimization path. When the maximum number of iterations is reached or the optimal path is stable, the current optimal path is output as the corrected speed size and speed direction of the first particle.