Dynamic evolution method for power distribution network structure

Abstract

The invention belongs to the technical field of data processing, and discloses a power distribution network structure dynamic evolution method, which comprises the following steps: constructing a topological structure model of a power system according to an actual structure of a power distribution network, and extracting feature information in spatio-temporal data by using a deep learning model based on the spatio-temporal data of the power distribution network to construct an adaptive feature library; the carbon emission optimization model is solved through the alternative feature library to obtain an alternative power distribution network structure, meanwhile, the actual adaptive feature library and the alternative feature library are compared in real time, adjustment is carried out by taking the most matched alternative power distribution network structure as a target, fine adjustment is carried out when necessary, on one hand, the real-time performance of adjustment is guaranteed, and on the other hand, the adjustment efficiency is improved; obvious delay caused by complex model calculation is avoided, the power distribution network roughly meets new requirements, and fine calculation is carried out when necessary to ensure that the power distribution network completely meets the requirements. According to the invention, low-delay and high-precision power distribution network structure adjustment is realized through two-order dynamic evolution.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a method for dynamic evolution of distribution network structure. Background Art

[0002] In the field of short-term power load forecasting, the commonly used technologies in the past include statistical models, traditional machine learning methods, and deep learning-based methods. Among them, the deep learning-based methods have achieved good results to a certain extent, especially the application of technologies such as convolutional neural networks (CNN) and bidirectional long short-term memory networks (BiLSTM). However, in the implementation and real-time adjustment links, although the existing technology may also have a feedback mechanism, the neural network model requires a lot of calculations from sensor data collection to the actual adjustment of the distribution network structure, and there may be a certain delay, resulting in untimely or inaccurate adjustments, affecting the carbon emission optimization effect. For example, when the load suddenly increases, due to feedback delays, it may not be possible to adjust the power generation distribution in time, resulting in grid instability or energy waste.

[0003] With the development of green and low-carbon concepts, solar and wind power generation devices have been added to many parts of the distribution network. The short-term fluctuations of these power generation devices are large. If the neural network model is required to perform calculations every time, the resulting delay may directly cause the adjustment of the distribution network to be completely out of sync with the actual situation, resulting in the adjustment action losing its practical significance and easily causing energy waste.

[0004] Therefore, how to achieve dynamic evolution of distribution network structure with low latency and high precision to avoid misalignment between distribution network adjustment and actual situation is a technical problem that needs to be solved urgently. Summary of the invention

[0005] In view of the problem that the information processing delay in the prior art is high, which leads to the misalignment between the adjustment of the distribution network and the actual situation, the present invention provides a method for the dynamic evolution of the distribution network structure. The carbon emission optimization model is solved in advance through the alternative feature library to obtain the alternative distribution network structure. At the same time, the actual adaptive feature library and the alternative feature library are compared in real time to adjust the best matching alternative distribution network structure, and fine-tune when necessary. On the one hand, the real-time nature of the adjustment is guaranteed, and obvious delays are avoided in complex model calculations, so that the distribution network roughly meets the new requirements. On the other hand, detailed calculations are performed when necessary to ensure that the distribution network fully meets the requirements. The present invention realizes low-latency and high-precision distribution network structure adjustment through two-order dynamic evolution.

[0006] The following is the technical solution of the present invention.

[0007] A method for dynamic evolution of distribution network structure, comprising the following steps: S1. Construct a topological structure model of the power system based on the actual structure of the distribution network; S2. Based on the spatiotemporal data of the distribution network, the deep learning model is used to extract the feature information in the spatiotemporal data and build an adaptive feature library; S3. Establish a carbon emission optimization model based on the adaptive feature library and topological structure model; S4. Adjust the variable features in the adaptive feature library to obtain an alternative feature library, solve the carbon emission optimization model using an optimization algorithm, obtain an alternative distribution network structure, repeat this step and establish a mapping relationship between several alternative feature libraries and corresponding alternative distribution network structures and store them; S5. Compare the actual adaptive feature library with the candidate feature library in real time. If the similarity is higher than the first threshold, perform the first dispatch: extract the alternative distribution network structure corresponding to the candidate feature library with the highest similarity, generate the first dispatch plan based on the difference between the current distribution network structure and the alternative distribution network structure, and execute the first dispatch; S6. Determine whether the highest similarity is higher than the second threshold. If not, perform the second scheduling: Based on the actual adaptive feature library, use the optimization algorithm to solve the carbon emission optimization model to obtain the optimal distribution network structure. Based on the difference between the distribution network structure after the first scheduling and the optimal distribution network structure, generate a second scheduling plan and execute the second scheduling.

[0008] In the present invention, the topological structure model shows the simplified structure of the distribution network, and the adaptive feature library covers the characteristic information that affects the distribution network. Adjusting the volatile features can be regarded as generating various plans. On this basis, the carbon emission optimization model is solved by the optimization algorithm to obtain the alternative distribution network structure, which can be used as a plan. In the actual operation process, due to the high frequency of change of the volatile features, the plan has covered the high probability of change, so it is easy to match similar situations, and the complexity of the similarity calculation is much smaller than the model calculation, so the first scheduling plan can be quickly obtained to meet the low latency requirements. After the first scheduling is completed, the distribution network has generally met the real-time demand, but the actual situation may also have deviations. At this time, the model can be further called for calculation to form an accurate second scheduling plan. Compared with the traditional direct call model, the present application uses the first scheduling plan to fill the delay time of the model calculation, match the current demand to the greatest extent, improve energy utilization, and reduce energy waste.

[0009] Preferably, the S1, constructing a topological structure model of the power system according to the actual structure of the distribution network, comprises: The actual structure of the distribution network is simplified by complex network theory, where nodes represent the various devices in the power system and edges represent the connection relationship between the devices. Among them, the connection relationship is represented by graph theory, and the degree k of each node is i represents the number of edges connected to the node, and the total degree of the graph It reflects the connectivity of the system, n is the number of nodes, and the connection relationship is expressed as: ; in, It is the element of the adjacency matrix A of the topological structure model.

[0010] Preferably, the S2, based on the spatiotemporal data of the distribution network, uses a deep learning model to extract feature information in the spatiotemporal data and construct an adaptive feature library, including: Use deep learning models built using temporal neural networks or graph convolutional networks to extract feature information from spatiotemporal data; Use data mining to further select and optimize the extracted feature information to obtain key features that affect carbon emissions and energy utilization; An adaptive feature library is constructed using key features, wherein the adaptive feature library is updated and adjusted in real time.

[0011] Preferably, the step S3, establishing a carbon emission optimization model based on an adaptive feature library and a topological structure model, comprises: The features in the adaptive feature library are combined with the power network topology model as comprehensive input data; Set the objective function: ,in, is the carbon emission coefficient of the ith node, x i is the power generation of the node, n is the number of nodes, and constraints including load balance are set to obtain the carbon emission optimization model.

[0012] Preferably, the step S4, adjusting the variable features in the adaptive feature library to obtain an alternative feature library, solving the carbon emission optimization model using an optimization algorithm to obtain an alternative distribution network structure, repeating this step and establishing and storing a mapping relationship between several alternative feature libraries and corresponding alternative distribution network structures, includes: Select several features with the most frequent fluctuations in the adaptive feature library as variable features, adjust the value of at least one variable feature each time, and obtain a new candidate feature library. Use the particle swarm optimization algorithm to solve the carbon emission optimization model and obtain the alternative distribution network structure, where the objective function of the particle swarm optimization algorithm is for: satisfy ; in, is the carbon emission coefficient of the ith node, x i is the power generation of the node, n is the number of nodes, and L is the load of the grid; Repeat the above steps, and establish and store mapping relationships between several candidate feature libraries and corresponding candidate distribution network structures.

[0013] Preferably, the S5, real-time comparison of the actual adaptive feature library and the alternative feature library, if the similarity is higher than the first threshold, then the first scheduling is performed: extracting the alternative distribution network structure corresponding to the alternative feature library with the highest similarity, generating a first scheduling plan based on the difference between the current distribution network structure and the alternative distribution network structure and executing the first scheduling, including: Taking each feature in the adaptive feature library and the candidate feature library as an element, an adaptive feature vector and a candidate feature vector are constructed respectively; Substitute the eigenvalues ​​of the actual adaptive feature library and the candidate feature library, calculate the cosine similarity between the adaptive feature vector and the candidate feature vector, and traverse different candidate feature libraries until the same number of cosine similarities as the candidate feature library is obtained; Determine whether there is a cosine similarity higher than the first threshold. If so, extract the alternative distribution network structure corresponding to the alternative feature library with the highest similarity, generate a first scheduling plan based on the difference between the current distribution network structure and the alternative distribution network structure, clarify the equipment, lines and loads that need to be adjusted, and the specific method of adjustment, and send the first scheduling plan to the scheduling execution system of the distribution network to execute the first scheduling.

[0014] Preferably, the S6, judging whether the highest similarity is higher than the second threshold, if not, performing the second scheduling: based on the actual adaptive feature library, solving the carbon emission optimization model by using the optimization algorithm, obtaining the optimal distribution network structure, generating the second scheduling plan based on the difference between the distribution network structure after the first scheduling and the optimal distribution network structure, and executing the second scheduling, including: It is determined whether the highest similarity is higher than the second threshold. If not, the particle swarm optimization algorithm is used to solve the carbon emission optimization model to obtain the optimal distribution network structure. The objective function of the particle swarm optimization algorithm is for: satisfy ; in, is the carbon emission coefficient of the ith node, x i is the power generation of the node, n is the number of nodes, and L is the load of the grid; Based on the difference between the distribution network structure after the first dispatch and the optimal distribution network structure, a second dispatch plan is generated, the equipment, lines and loads that need to be adjusted, as well as the specific method of adjustment are clearly defined, and the second dispatch plan is sent to the dispatch execution system of the distribution network to execute the second dispatch.

[0015] Preferably, the data mining includes K-means clustering method or association rule mining method.

[0016] The present invention also provides an electronic 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 for dynamic evolution of distribution network structure when calling the computer program in the memory.

[0017] The present invention also provides a storage medium, in which computer executable instructions are stored. When the computer executable instructions are loaded and executed by a processor, the steps of the above-mentioned method for dynamic evolution of distribution network structure are implemented.

[0018] The substantial effects of the present invention include: First, we use complex network theory to simplify the actual structure of the distribution network and build a topological structure model, with nodes representing the various devices in the power system and edges representing the connection relationship between the devices. Then, based on the spatiotemporal data of the distribution network, we use a deep learning model to extract feature information, build an adaptive feature library, and update and adjust it in real time. Then, we combine the adaptive feature library and the topological structure model to establish a carbon emission optimization model with the goal of minimizing carbon emissions.

[0019] During the implementation of the method, the alternative feature library is obtained by adjusting the mutable features in the adaptive feature library, and the carbon emission optimization model is solved by the optimization algorithm to obtain the alternative distribution network structure. The actual adaptive feature library and the alternative feature library are compared in real time. If the similarity is higher than the set threshold, the corresponding alternative distribution network structure is directly extracted for the first dispatch, which quickly responds to the changing needs of the distribution network and avoids the delay caused by complex model calculations. At the same time, detailed calculations are performed when necessary to ensure that the distribution network structure fully meets actual needs, improve energy utilization, and reduce energy waste. In addition, the method also has good adaptability and scalability, and can cope with the short-term volatility of power generation devices such as solar energy and wind energy in the distribution network, providing strong technical support for the intelligent management and optimization of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution will be clearly and completely described below in combination with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] It should be understood that in various embodiments of the present invention, the size of the sequence number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0023] It should be understood that in the present invention, "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising 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 these processes, methods, products or apparatuses.

[0024] It should be understood that in the present invention, "plurality" refers to two or more than two. "And / or" is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "Contains A, B and C", "Contains A, B, C" means that A, B, and C are all included, "Contains A, B or C" means that one of A, B, and C is included, and "Contains A, B and / or C" means that any one, any two, or any three of A, B, and C are included.

[0025] The technical solution of the present invention is described in detail with specific embodiments below. The embodiments may be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0026] Example: A method for dynamic evolution of distribution network structure, comprising: Figure 1 The following steps are shown: S1. Construct a topological model of the power system based on the actual structure of the distribution network.

[0027] In this embodiment, S1 includes: The actual structure of the distribution network is simplified by complex network theory, where nodes represent the various devices in the power system and edges represent the connection relationship between the devices. Among them, the connection relationship is represented by graph theory, and the degree k of each node is i represents the number of edges connected to the node, and the total degree of the graph It reflects the connectivity of the system, n is the number of nodes, and the connection relationship is expressed as: ; in, It is the element of the adjacency matrix A of the topological structure model.

[0028] For example, substations and various equipment can be represented by nodes.

[0029] S2. Based on the spatiotemporal data of the distribution network, a deep learning model is used to extract feature information from the spatiotemporal data and build an adaptive feature library.

[0030] In this embodiment, S2 includes: Use deep learning models built using temporal neural networks or graph convolutional networks to extract feature information from spatiotemporal data; Use data mining to further select and optimize the extracted feature information to obtain key features that affect carbon emissions and energy utilization; An adaptive feature library is constructed using key features, wherein the adaptive feature library is updated and adjusted in real time.

[0031] Among them, spatiotemporal data can be collected in real time through smart meters, sensors, weather stations and other equipment, including load data, equipment status data, meteorological data, etc. Missing data is filled by interpolation methods or mean filling, and load data and other continuous values ​​are standardized to ensure that the data is suitable for the input of the neural network. The spatiotemporal data is then divided according to the time window to prepare for subsequent model training.

[0032] If a temporal neural network is used, the basic formula is: ; in, is the time step The hidden state of are inputs (such as historical loads, weather data, etc.), and is the weight matrix, is the bias term.

[0033] If a graph convolutional network is used, the formula is: ; in, is the node feature matrix of the lth layer, A' is the normalized adjacency matrix, is the weight matrix of the layer and σ is the activation function.

[0034] In this embodiment, the data mining includes K-means clustering method or association rule mining method.

[0035] The objective function of the K-means clustering method is to minimize the distance from each point to the center of the cluster to which it belongs: ; Where m is the number of samples, K is the number of clusters, and μ k is the center of cluster k, is the i-th sample.

[0036] Support in association rule mining: P(A∩B)≥δ; Among them, A and B are item sets, P(A∩B) is the support, and δ is the minimum support threshold.

[0037] Through K-means clustering method or association rule mining method, the extracted feature information can be further selected and optimized to obtain the key features that affect carbon emissions and energy utilization.

[0038] S3. Establish a carbon emission optimization model based on the adaptive feature library and topological structure model.

[0039] In this embodiment, S3 includes: The features in the adaptive feature library are combined with the power network topology model as comprehensive input data; Set the objective function: ,in, is the carbon emission coefficient of the ith node, x i is the power generation of the node, n is the number of nodes, and constraints including load balance are set to obtain the carbon emission optimization model.

[0040] S4. Adjust the variable features in the adaptive feature library to obtain an alternative feature library, use the optimization algorithm to solve the carbon emission optimization model, and obtain an alternative distribution network structure. Repeat this step and establish a mapping relationship between several alternative feature libraries and corresponding alternative distribution network structures and store them.

[0041] In this embodiment, S4 includes: Select several features with the most frequent fluctuations in the adaptive feature library as variable features, and adjust the value of at least one variable feature each time to obtain a new candidate feature library.

[0042] For example, among the several features of the adaptive feature library of a distribution network, feature 3 (real-time load demand) and feature 5 (photovoltaic power generation) fluctuate most frequently and have a more obvious impact on the adjustment decision of the distribution network structure. These two features are selected as variable features.

[0043] For example, the real-time load demand in the current adaptive feature library is 100MW, and the photovoltaic power generation is 20MWh. In order to generate a new candidate feature library, the value of at least one variable feature is adjusted each time. For example: The first adjustment: increase the real-time load demand by 10MW to 110MW, and keep the photovoltaic power generation unchanged.

[0044] Second adjustment: Increase photovoltaic power generation by 5MWh to 25MWh, and keep real-time load demand unchanged.

[0045] The third adjustment: adjust the real-time load demand and photovoltaic power generation at the same time, increase the real-time load demand by 5MW to 105MW, and increase the photovoltaic power generation by 3MWh to 23MWh.

[0046] After the above adjustments, three new candidate feature libraries can be obtained, in which the variable features correspond to one of the above adjustment results.

[0047] For each new candidate feature library, the particle swarm optimization algorithm is used to solve the carbon emission optimization model to obtain the alternative distribution network structure, where the objective function of the particle swarm optimization algorithm is for: satisfy ; in, is the carbon emission coefficient of the ith node, x i is the power generation of the node, n is the number of nodes, and L is the load of the grid; Repeat the above steps, and establish and store mapping relationships between several candidate feature libraries and corresponding candidate distribution network structures.

[0048] S5. Compare the actual adaptive feature library with the alternative feature library in real time. If the similarity is higher than the first threshold, perform the first scheduling: extract the alternative distribution network structure corresponding to the alternative feature library with the highest similarity, generate the first scheduling plan based on the difference between the current distribution network structure and the alternative distribution network structure, and execute the first scheduling.

[0049] In this embodiment, S5 includes: Taking each feature in the adaptive feature library and the candidate feature library as an element, an adaptive feature vector and a candidate feature vector are constructed respectively; Substitute the eigenvalues ​​of the actual adaptive feature library and the candidate feature library, calculate the cosine similarity between the adaptive feature vector and the candidate feature vector, and traverse different candidate feature libraries until the same number of cosine similarities as the candidate feature library is obtained; Determine whether there is a cosine similarity higher than the first threshold. If so, extract the alternative distribution network structure corresponding to the alternative feature library with the highest similarity, generate a first scheduling plan based on the difference between the current distribution network structure and the alternative distribution network structure, clarify the equipment, lines and loads that need to be adjusted, and the specific method of adjustment, and send the first scheduling plan to the scheduling execution system of the distribution network to execute the first scheduling.

[0050] For the sake of convenience, a simplified adaptive feature library is taken as an example. For example, the adaptive feature library contains 5 features. The adaptive feature vector of the current distribution network is [130, 320, 68, 18, 110], the candidate feature vector A = [120,300, 65, 15, 120], and the candidate feature vector B = [150, 350, 70, 20, 90]. The cosine similarity of the adaptive feature vector and the candidate feature vector is calculated respectively, and the cosine similarity cos(θA)=0.92 and the cosine similarity cos(θB)=0.95 are obtained. If the first threshold is 0.9, there is a similarity higher than the first threshold, and the candidate feature library with the highest similarity is the candidate feature library B, then the corresponding alternative distribution network structure is extracted, and the first scheduling plan is generated based on the difference between the current distribution network structure and the alternative distribution network structure.

[0051] Of course, when necessary, each dimension of the feature vector can be normalized before calculating the cosine similarity.

[0052] S6. Determine whether the highest similarity is higher than the second threshold. If not, perform the second scheduling: Based on the actual adaptive feature library, use the optimization algorithm to solve the carbon emission optimization model to obtain the optimal distribution network structure. Based on the difference between the distribution network structure after the first scheduling and the optimal distribution network structure, generate a second scheduling plan and execute the second scheduling.

[0053] In this embodiment, S6 includes: It is determined whether the highest similarity is higher than the second threshold. If not, the particle swarm optimization algorithm is used to solve the carbon emission optimization model to obtain the optimal distribution network structure. The objective function of the particle swarm optimization algorithm is for: satisfy ; in, is the carbon emission coefficient of the ith node, x i is the power generation of the node, n is the number of nodes, and L is the load of the grid; Based on the difference between the distribution network structure after the first dispatch and the optimal distribution network structure, a second dispatch plan is generated, the equipment, lines and loads that need to be adjusted, as well as the specific method of adjustment are clearly defined, and the second dispatch plan is sent to the dispatch execution system of the distribution network to execute the second dispatch.

[0054] It should be noted that the second threshold is higher than the first threshold, for example, 0.95. When it is higher than the second threshold, it can be considered that the actual situation is basically the same as the situation of the backup feature library, and the scheduling result is considered to be in line with the current situation. If it is not higher than the second threshold, it is considered that there is still room for fine-tuning.

[0055] In this embodiment, the topological structure model shows the simplified structure of the distribution network, and the adaptive feature library covers the characteristic information that affects the distribution network. Adjusting the volatile features can be regarded as generating various plans. On this basis, the carbon emission optimization model is solved by the optimization algorithm to obtain the alternative distribution network structure, which can be used as a plan. In the actual operation process, due to the high frequency of change of the volatile features, the plan has covered the high probability of change, so it is easy to match similar situations, and the complexity of the similarity calculation is much smaller than the model calculation, so the first scheduling plan can be quickly obtained to meet the low latency requirements. After the first scheduling is completed, the distribution network has generally met the real-time demand, but the actual situation may also have deviations. At this time, the model can be further called for calculation to form an accurate second scheduling plan. Compared with the traditional direct call model, this application uses the first scheduling plan to fill the delay time of the model calculation, match the current demand to the greatest extent, improve energy utilization, and reduce energy waste.

[0056] This embodiment also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, the steps of the above-mentioned method for dynamic evolution of distribution network structure are implemented.

[0057] This embodiment further provides a storage medium, in which computer executable instructions are stored. When the computer executable instructions are loaded and executed by a processor, the steps of the above-mentioned method for dynamic evolution of distribution network structure are implemented.

[0058] In summary, this embodiment first uses complex network theory to simplify the actual structure of the distribution network and constructs a topological structure model, with nodes representing the various devices in the power system and edges representing the connection relationship between the devices. Then, based on the spatiotemporal data of the distribution network, a deep learning model is used to extract feature information, build an adaptive feature library, and update and adjust it in real time. Then, a carbon emission optimization model is established by combining the adaptive feature library and the topological structure model, with the goal of minimizing carbon emissions.

[0059] During the implementation process, the alternative feature library is obtained by adjusting the mutable features in the adaptive feature library, and the carbon emission optimization model is solved by the optimization algorithm to obtain the alternative distribution network structure. The actual adaptive feature library and the alternative feature library are compared in real time. If the similarity is higher than the set threshold, the corresponding alternative distribution network structure is directly extracted for the first dispatch, which quickly responds to the changing needs of the distribution network and avoids the delay caused by complex model calculations. At the same time, detailed calculations are performed when necessary to ensure that the distribution network structure fully meets actual needs, improve energy utilization, and reduce energy waste. In addition, this method also has good adaptability and scalability, and can cope with the short-term volatility of power generation devices such as solar energy and wind energy in the distribution network, providing strong technical support for the intelligent management and optimization of the distribution network.

[0060] Through the description of the above implementation methods, technical personnel in the relevant field can understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the specific device can be divided into different functional modules to complete all or part of the functions described above.

[0061] In the embodiments provided in the present application, it should be understood that the disclosed structures and methods can be implemented in other ways. For example, the embodiments of the structure described above are only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another structure, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, structures or units, which can be electrical, mechanical or other forms.

[0062] The units described as separate components may or may not be physically separated, and the components shown as units may be one physical unit or multiple physical units, that is, they may be located in one place or distributed in multiple different places. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0063] In addition, each functional unit in the embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0064] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application is essentially 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, which is stored in a storage medium, including several instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) to perform all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0065] The above contents are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for dynamic evolution of distribution network structure, characterized in that: The following steps are involved: S1. Construct a topological structure model of the power system based on the actual structure of the distribution network; S2. Based on the spatiotemporal data of the distribution network, the deep learning model is used to extract the feature information in the spatiotemporal data and build an adaptive feature library; S3. Establish a carbon emission optimization model based on the adaptive feature library and topological structure model; S4. Adjust the variable features in the adaptive feature library to obtain an alternative feature library, solve the carbon emission optimization model using an optimization algorithm, obtain an alternative distribution network structure, repeat this step and establish a mapping relationship between several alternative feature libraries and corresponding alternative distribution network structures and store them; S5. Compare the actual adaptive feature library with the candidate feature library in real time. If the similarity is higher than the first threshold, perform the first dispatch: extract the alternative distribution network structure corresponding to the candidate feature library with the highest similarity, generate the first dispatch plan based on the difference between the current distribution network structure and the alternative distribution network structure, and execute the first dispatch; S6. Determine whether the highest similarity is higher than the second threshold. If not, perform the second scheduling: Based on the actual adaptive feature library, use the optimization algorithm to solve the carbon emission optimization model to obtain the optimal distribution network structure. Based on the difference between the distribution network structure after the first scheduling and the optimal distribution network structure, generate a second scheduling plan and execute the second scheduling.

2. A method for dynamic evolution of distribution network structure according to claim 1, characterized in that: S1, constructing a topological structure model of the power system according to the actual structure of the distribution network, including: The actual structure of the distribution network is simplified by complex network theory, where nodes represent the various devices in the power system and edges represent the connection relationship between the devices. Among them, the connection relationship is represented by graph theory, and the degree of each node represents the number of edges connected to the node, and the total degree of the graph It reflects the connectivity of the system, n is the number of nodes, and the connection relationship is expressed as: ； in, It is the element of the adjacency matrix A of the topological structure model.

3. A method for dynamic evolution of distribution network structure according to claim 1, characterized in that: S2, based on the spatiotemporal data of the distribution network, using a deep learning model to extract feature information in the spatiotemporal data and construct an adaptive feature library, including: Use deep learning models built using temporal neural networks or graph convolutional networks to extract feature information from spatiotemporal data; Use data mining to further select and optimize the extracted feature information to obtain key features that affect carbon emissions and energy utilization; An adaptive feature library is constructed using key features, wherein the adaptive feature library is updated and adjusted in real time.

4. A method for dynamic evolution of distribution network structure according to claim 1, characterized in that: S3, based on the adaptive feature library and the topological structure model, establishes a carbon emission optimization model, including: The features in the adaptive feature library are combined with the power network topology model as comprehensive input data; Set the objective function: ,in, is the carbon emission coefficient of the ith node, x i is the power generation of the node, n is the number of nodes, and constraints including load balance are set to obtain the carbon emission optimization model.

5. A method for dynamic evolution of distribution network structure according to claim 4, characterized in that: The step S4, adjusting the variable features in the adaptive feature library to obtain an alternative feature library, solving the carbon emission optimization model using an optimization algorithm to obtain an alternative distribution network structure, repeating this step and establishing and storing a mapping relationship between several alternative feature libraries and corresponding alternative distribution network structures, includes: Select several features with the most frequent fluctuations in the adaptive feature library as variable features, adjust the value of at least one variable feature each time, and obtain a new candidate feature library. Use the particle swarm optimization algorithm to solve the carbon emission optimization model and obtain the alternative distribution network structure, where the objective function of the particle swarm optimization algorithm is for: satisfy ; in, is the carbon emission coefficient of the ith node, x i is the power generation of the node, n is the number of nodes, and L is the load of the grid; Repeat the above steps, and establish and store mapping relationships between several candidate feature libraries and corresponding candidate distribution network structures.

6. A method for dynamic evolution of distribution network structure according to claim 1, characterized in that: The S5, comparing the actual adaptive feature library with the candidate feature library in real time, if the similarity is higher than the first threshold, performing the first dispatch: extracting the candidate distribution network structure corresponding to the candidate feature library with the highest similarity, generating the first dispatch plan based on the difference between the current distribution network structure and the candidate distribution network structure, and executing the first dispatch, includes: Taking each feature in the adaptive feature library and the candidate feature library as an element, an adaptive feature vector and a candidate feature vector are constructed respectively; Substitute the eigenvalues ​​of the actual adaptive feature library and the candidate feature library, calculate the cosine similarity between the adaptive feature vector and the candidate feature vector, and traverse different candidate feature libraries until the same number of cosine similarities as the candidate feature library is obtained; Determine whether there is a cosine similarity higher than the first threshold. If so, extract the alternative distribution network structure corresponding to the alternative feature library with the highest similarity, generate a first scheduling plan based on the difference between the current distribution network structure and the alternative distribution network structure, clarify the equipment, lines and loads that need to be adjusted, and the specific method of adjustment, and send the first scheduling plan to the scheduling execution system of the distribution network to execute the first scheduling.

7. A method for dynamic evolution of distribution network structure according to claim 6, characterized in that: S6, judging whether the highest similarity is higher than the second threshold, if not, performing the second dispatch: based on the actual adaptive feature library, using the optimization algorithm to solve the carbon emission optimization model, obtaining the optimal distribution network structure, generating the second dispatch plan based on the difference between the distribution network structure after the first dispatch and the optimal distribution network structure, and executing the second dispatch, including: It is determined whether the highest similarity is higher than the second threshold. If not, the particle swarm optimization algorithm is used to solve the carbon emission optimization model to obtain the optimal distribution network structure. The objective function of the particle swarm optimization algorithm is for: satisfy ; in, is the carbon emission coefficient of the ith node, x i is the power generation of the node, n is the number of nodes, and L is the load of the grid; Based on the difference between the distribution network structure after the first dispatch and the optimal distribution network structure, a second dispatch plan is generated, the equipment, lines and loads that need to be adjusted, as well as the specific method of adjustment are clearly defined, and the second dispatch plan is sent to the dispatch execution system of the distribution network to execute the second dispatch.

8. A method for dynamic evolution of distribution network structure according to claim 3, characterized in that: The data mining includes K-means clustering method or association rule mining method.

9. An electronic device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, it implements the steps of a distribution network structure dynamic evolution method as claimed in any one of claims 1 to 8.

10. A storage medium, characterized in that: The storage medium stores computer executable instructions, and when the computer executable instructions are loaded and executed by the processor, the steps of a distribution network structure dynamic evolution method as described in any one of claims 1 to 8 are implemented.

Images