Power distribution network dynamic reconfiguration method, device, equipment, storage medium and program product
By acquiring multi-regional data from the distribution network and performing federated learning and reinforcement learning, the topology structure is optimized, solving the problem of low accuracy in distribution network reconfiguration caused by a single data source, and achieving more efficient dynamic reconfiguration and energy allocation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- MEIZHOU POWER SUPPLY BUREAU OF GUANGDONG POWER GRID CORP
- Filing Date
- 2026-01-15
- Publication Date
- 2026-05-29
Smart Images
Figure CN122118764A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of electrical technology, and in particular to a method, apparatus, equipment, storage medium, and program product for dynamic reconfiguration of power distribution networks. Background Technology
[0002] With the rapid development of smart grids, dynamic reconfiguration of distribution networks plays a crucial role in improving power supply reliability, optimizing energy distribution, and enhancing system stability.
[0003] Related technologies rely on a single data source of the distribution network (such as load data alone) for data analysis and decision-making. Furthermore, there are data barriers between different geographical areas of the distribution network, and different geographical areas often make independent decisions to determine the dynamic reconfiguration results of the distribution network in each geographical area.
[0004] Based on the above analysis, it can be concluded that the relevant technologies suffer from low accuracy in dynamic reconfiguration of power distribution networks due to independent decision-making between different geographical regions and the use of only a single data source for analysis and decision-making. Summary of the Invention
[0005] This application provides a method, apparatus, equipment, storage medium, and program product for dynamic reconfiguration of distribution networks, in order to improve the accuracy of dynamic reconfiguration structures of distribution networks.
[0006] In a first aspect, embodiments of this application provide a method for dynamic reconfiguration of a power distribution network, comprising: acquiring load data, equipment status data, and environmental data of multiple power distribution devices in the power distribution network; wherein the multiple power distribution devices are located in different geographical areas;
[0007] The load data, equipment status data, and environmental data of each of the multiple substation devices are input into a pre-trained target prediction model. The target prediction model then predicts the deviation values of the preset operating indicators of each of the multiple substation devices, thereby obtaining the deviation values of the preset operating indicators of each of the multiple substation devices. The target prediction model is obtained through federated learning of the historical load data, historical equipment status data, historical environmental data of multiple substation devices in different geographical areas, and the historical deviation data of the preset operating indicators.
[0008] The topology of the power distribution network is updated based on the deviation values of the preset operating indicators of the multiple power equipment.
[0009] The distribution network is reconstructed based on the updated topology to obtain the reconstructed distribution network.
[0010] In one possible implementation, updating the topology of the distribution network based on the deviation values of the preset operating indicators of the plurality of power equipment includes:
[0011] The deviation values of the preset operating indicators of each of the multiple power equipment and the topology of the distribution network are input into the decision model, and the decision model outputs the updated topology of the distribution network.
[0012] In one possible implementation, the updated distribution network topology is determined by the decision model based on the following steps:
[0013] Based on the deviation value, load adjustment data for multiple power equipment are determined;
[0014] Based on the load adjustment data of multiple power equipment, determine the load adjustment data of power equipment spanning geographical regions;
[0015] Based on the load adjustment data of the cross-geographical substations and to minimize the total active power loss of the distribution network, the topology of the updated distribution network is determined.
[0016] In one possible implementation, the target prediction model is obtained by performing the following federated learning operation based on the local servers and central servers corresponding to each geographical region:
[0017] For each geographic region, the local server corresponding to that geographic region learns the gradient information of the local prediction model corresponding to that geographic region based on the historical data of the power equipment in that geographic region.
[0018] Each of the local servers encrypts its own gradient information and uploads it to the central server;
[0019] The central server decrypts the gradient information corresponding to each of the local servers, aggregates it, and determines the global prediction model based on the aggregation result.
[0020] In response to the failure to meet the preset convergence condition, the local prediction models corresponding to each geographical region are updated using the global prediction model, and the federated learning operation is re-executed.
[0021] In response to satisfying the preset convergence condition, the global prediction model that satisfies the preset convergence condition is taken as the target prediction model.
[0022] In one possible implementation, the step of learning the gradient information of the local prediction model corresponding to each geographical region by a local server based on historical data of the substations in that geographical region includes:
[0023] For each geographical region, the local server corresponding to that geographical region trains a local prediction model for that geographical region based on the historical load data, historical equipment status data, historical environmental data, and historical deviation data of preset operating indicators of each power equipment in that geographical region; and obtains the gradient information of the local prediction model for that geographical region.
[0024] The gradient information of the local prediction model includes the influence weights of load data, equipment status data, and environmental data on the deviation values of preset operating indicators.
[0025] In one possible implementation, the central server decrypts the gradient information corresponding to each of the local servers, aggregates it, and determines a global prediction model based on the aggregation result, including:
[0026] The central server aggregates the gradient information of the local prediction models corresponding to multiple geographical regions based on a preset aggregation algorithm to obtain aggregated gradient information; the aggregated gradient information includes aggregated data after aggregating the influence weights corresponding to the load data, the equipment status data, and the environmental data;
[0027] The global prediction model is determined based on the aggregated data corresponding to the load data, the equipment status data, and the environmental data.
[0028] In one possible implementation, the central server aggregates the gradient information of the local prediction models corresponding to multiple geographical regions based on a preset aggregation algorithm to obtain aggregated gradient information, including:
[0029] The central server determines the aggregation weight for each geographical region based on the amount and quality of historical load data, historical equipment status data, and historical environmental data of each power equipment in each geographical region.
[0030] For each data point in the load data, equipment status data, and environmental data, the weighted average of the influence weights of that data is calculated based on the aggregation weights of each geographical region, thus obtaining the aggregated data corresponding to that data.
[0031] In one possible implementation, the operating index data of the distribution network includes at least one of the following: line loss and voltage.
[0032] In one possible implementation, updating the topology of the distribution network based on the load adjustment data of the cross-geographical area substations and to minimize the total active power loss of the distribution network includes:
[0033] Determine the first state of multiple switches in the topology of the power distribution network, wherein each switch controls the on / off state of the power transmission line between two substations in the topology of the power distribution network.
[0034] The topology of the distribution network having a first state of multiple switches is input into a preset reinforcement learning model, and the reinforcement learning model outputs a second state of multiple switches; wherein the second state of the multiple second switches minimizes the total active power loss of the distribution network.
[0035] Update the first state of multiple switches to their respective second states.
[0036] Secondly, embodiments of this application provide a power distribution network dynamic reconfiguration device, comprising:
[0037] The acquisition module is used to acquire load data, equipment status data, and environmental data of multiple substations in the power distribution network; wherein the multiple substations are located in different geographical areas.
[0038] The prediction module is used to input the load data, equipment status data, and environmental data of each of the multiple substation devices into a pre-trained target prediction model. The target prediction model then predicts the deviation values of the preset operating indicators of each of the multiple substation devices to obtain the deviation values of the preset operating indicators of each of the multiple substation devices. The target prediction model is obtained by federated learning from the historical load data, historical equipment status data, historical environmental data of multiple substation devices in different geographical areas, as well as the historical deviation data of the preset operating indicators.
[0039] The update module is used to update the topology of the power distribution network based on the deviation values of the preset operating indicators of the multiple power equipment.
[0040] The reconfiguration module is used to reconfigure the distribution network according to the updated topology to obtain the reconfigured distribution network.
[0041] Thirdly, embodiments of this application provide a power distribution network dynamic reconfiguration device, including: a memory and a processor;
[0042] The memory stores computer-executed instructions;
[0043] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.
[0044] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.
[0045] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.
[0046] The distribution network dynamic reconfiguration method, apparatus, equipment, storage medium, and program product provided in this application acquire load data, equipment status data, and environmental data of substation equipment in different geographical regions of the distribution network. They then process this data using a predictive model to obtain deviation values of preset operating indicators for each substation. Based on these deviation values, the distribution network topology is updated, and dynamic reconfiguration is performed according to the updated topology. By analyzing the impact of various cross-regional data on the distribution network and performing dynamic reconfiguration based on the analysis results, the accuracy of dynamic distribution network reconfiguration is improved. Attached Figure Description
[0047] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0048] Figure 1 A flowchart illustrating the dynamic reconfiguration method for power distribution networks provided in this application embodiment;
[0049] Figure 2 This is a schematic diagram illustrating the process of updating the topology of a distribution network as provided in an embodiment of this application.
[0050] Figure 3 A schematic diagram illustrating the learning operation of the prediction model provided in the embodiments of this application;
[0051] Figure 4 This is a schematic diagram of the structure of the power distribution network dynamic reconfiguration device provided in the embodiments of this application;
[0052] Figure 5 This is a schematic diagram of the structure of the power distribution network dynamic reconfiguration device provided in the embodiments of this application.
[0053] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation
[0054] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.
[0055] With the rapid development of smart grids, dynamic reconfiguration of distribution networks plays a crucial role in improving power supply reliability, optimizing energy distribution, and enhancing system stability. Reconfiguration of distribution networks using related technologies primarily relies on analysis and decision-making based on a single data source (such as load data). However, this single-data-driven approach struggles to comprehensively reflect the actual operating status of the distribution network. In complex operating environments, these technologies often lead to biased and inefficient reconfiguration decisions, failing to meet the real-time and accuracy requirements of modern distribution networks.
[0056] Furthermore, the relevant technologies have significant shortcomings in cross-geographical collaborative optimization. Data between different geographical regions is often not shared due to privacy protection and security barriers, making it difficult for the relevant technologies to achieve global optimization and severely restricting the efficient operation and collaborative management capabilities of the power distribution network in complex scenarios.
[0057] Based on the above analysis, it can be concluded that when related technologies restructure the distribution network, the reliance on a single data source for analysis and decision-making, coupled with data barriers between different geographical regions, leads to technical problems in the accuracy of dynamic restructuring of the distribution network.
[0058] The dynamic reconfiguration method for distribution networks provided in this application is used to solve the above-mentioned technical problems.
[0059] The technical solution of this application and how it solves the above-mentioned technical problems will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.
[0060] Figure 1 This is a flowchart illustrating the dynamic reconfiguration method for distribution networks provided in the embodiments of this application, as shown below. Figure 1 As shown, the method includes:
[0061] S101. Obtain the load data, equipment status data, and environmental data of multiple substations in the power distribution network; wherein, the multiple substations are located in different geographical areas.
[0062] It should be noted that load data, equipment status data, and environmental data affect the operation of the distribution network from different perspectives, and these data can be various types of data over a period of time. For example, load data includes the load current, power, and energy of the substation equipment in the distribution network; equipment status data includes the oil temperature of the transformer and the current of the lines in the substation equipment; and environmental data includes the temperature, humidity, and wind speed of the environment in which the substation equipment is located.
[0063] Raw data from various power distribution equipment in different geographical areas can be collected using multiple sensors. To improve data quality and ensure data consistency, the extracted raw data can be preprocessed. For example, the extracted raw data can be normalized and denoised to obtain processed data that can be used as input load data, equipment status data, and environmental data.
[0064] S102. Input the load data, equipment status data and environmental data of each of the multiple substation devices into the pre-trained target prediction model. The target prediction model predicts the deviation values of the preset operating indicators of each substation device, and obtains the deviation values of the preset operating indicators of each of the multiple substation devices. The target prediction model is obtained by federated learning of the historical load data, historical equipment status data and historical environmental data of multiple substation devices in different geographical areas and the historical deviation data of the preset operating indicators.
[0065] It should be noted that preset operating index data can be used to determine the operating status of power equipment. For example, the operating index data of the distribution network includes at least one of the following: line loss and voltage.
[0066] S103. Update the topology of the distribution network based on the deviation values of the preset operating indicators of multiple power equipment.
[0067] S104. Reconstruct the distribution network according to the updated topology to obtain the reconstructed distribution network.
[0068] As an example, control commands for power distribution equipment can be generated based on the updated topology. By issuing these control commands to the power distribution equipment, the equipment can adjust its operating status accordingly, thereby achieving dynamic reconfiguration of the power distribution network.
[0069] The dynamic reconfiguration method for distribution networks provided in this application acquires load data, equipment status data, and environmental data of substation equipment in different geographical regions of the distribution network. It then processes this data using a predictive model to obtain deviation values of preset operating indicators for each substation. Based on these deviation values, the topology of the distribution network is updated, and dynamic reconfiguration is performed according to the updated topology. By analyzing the impact of various cross-regional data on the distribution network and performing dynamic reconfiguration based on the analysis results, the accuracy of dynamic reconfiguration of the distribution network is improved.
[0070] In some specific embodiments, S103 above includes the following steps:
[0071] The deviation values of the preset operating indicators of multiple power equipment and the topology of the distribution network are input into the decision model, and the decision model outputs the updated topology of the distribution network.
[0072] It should be noted that the input distribution network topology includes the topology of multiple geographical regions. The updated distribution network topology output by the decision model, based on the deviation values of the preset operating indicators of multiple substations and the distribution network topology, also includes the topology of multiple geographical regions. The topology includes multiple substations within these geographical regions and the electrical connections between them. The updated distribution network topology is adjusted based on the deviation values of the preset operating indicators of the multiple substations, thereby optimizing energy distribution and ensuring the stable operation of the distribution network.
[0073] For example, a topology includes multiple nodes and the connections between them. Each node includes node information, such as equipment information (e.g., location and electrical information), and the connections between nodes characterize the electrical connections between them. The topology can be changed by altering the electrical connections between nodes.
[0074] Figure 2 This is a schematic diagram of the process for updating the topology of the distribution network provided in the embodiments of this application, such as... Figure 2 As shown, in some implementations of these embodiments, the updated distribution network topology is determined by a decision model based on the following steps:
[0075] S201. Determine the load adjustment data for multiple power equipment based on the deviation value.
[0076] For example, if the decision model detects that "the load of multiple power substations in a certain geographical area has increased by 10% under high temperature weather" based on the input data, and outputs "the line loss of multiple power substations in the geographical area has increased by 5%", then the load adjustment data of each power substation can be determined based on the output line loss increase value and the operating status of the multiple power substations.
[0077] S202. Based on the load adjustment data of multiple power equipment, determine the load adjustment data of power equipment that crosses geographical regions.
[0078] For example, if the problem of line loss cannot be solved by adjusting the load of the multiple power substations, the load of power substations in other geographical areas can be adjusted in coordination to solve the problem of line loss.
[0079] S203. Based on the load adjustment data of cross-geographical substations and to minimize the total active power loss of the distribution network, determine the updated distribution network topology.
[0080] In some implementations of these embodiments, S203 above includes the following steps:
[0081] First, determine the first state of multiple switches in the topology of the distribution network, wherein each switch controls the on / off state of the power transmission line between two substations in the topology of the distribution network.
[0082] It should be noted that the node information of the distribution network topology includes substation equipment information and switch information, and the first state of the switch can be obtained through the node information.
[0083] Second, the topology of the distribution network with multiple switches in the first state is input into a preset reinforcement learning model, and the reinforcement learning model outputs the second state of multiple switches; among them, the second state of multiple second switches minimizes the total active power loss of the distribution network.
[0084] It should be noted that the pre-trained reinforcement learning model can be a reinforcement learning model trained based on historical data of the distribution network. The trained model can output a second state of multiple switches based on the current state of multiple switches in the distribution network, and this second state minimizes the total active power loss of the distribution network.
[0085] For the training process of reinforcement learning models, for example, a deep Q-network algorithm can be used to train the model, and the model training update formula can be expressed by the following formula (1):
[0086] (1);
[0087] In the formula, Q(s,a) represents the value function of taking action a to open or close a switch under the switching states s of multiple substations. This value function can be used to evaluate the total active power loss of the distribution network; α represents the learning rate; r represents the immediate reward; γ represents the discount factor; s′ represents the next state; and a′ represents the next action.
[0088] By using historical data on the switching status of multiple substations in the power distribution network, the model is iteratively updated based on the above formula (1) to determine the final reinforcement learning model.
[0089] Third, update the first state of multiple switches to their respective second states.
[0090] In these implementations, load adjustment data for multiple power distribution devices is determined based on deviation values. Then, a reinforcement learning model is used to determine the state of the switches corresponding to the multiple power distribution devices based on the load adjustment data. The switch states are adjusted according to the determined switch states, thereby changing the line connectivity in the distribution network and redistributing the load of the distribution network to optimize energy distribution.
[0091] Figure 3 A flowchart illustrating the learning operation of the prediction model provided in the embodiments of this application is shown below. Figure 3 As shown, the target prediction model is obtained by performing the following federated learning operations on the local servers and central servers corresponding to each geographical region:
[0092] S301. For each geographic region, the local server corresponding to that geographic region learns the gradient information of the local prediction model corresponding to that geographic region based on the historical data of the substation equipment in that geographic region.
[0093] In some implementations of these embodiments, S301 includes: for each geographical region, a local server corresponding to that geographical region trains a local prediction model for that geographical region based on historical load data, historical equipment status data, historical environmental data, and historical deviation data of preset operating indicators for each substation within that geographical region; and obtains the gradient information of the local prediction model for that geographical region. The gradient information of the local prediction model includes the influence weights of load data, equipment status data, and environmental data on the deviation values of the preset operating indicators.
[0094] It should be noted that the training operations of the local servers run independently, and the historical data used for training on the local servers in different regions are obtained independently from each other. The amount of historical data obtained can be determined based on the data acquisition devices in the region and the acquisition strategy controlled independently by each region.
[0095] For the gradient information of the above local prediction model, the larger the influence weight value in the gradient information, the greater the influence of the corresponding data on the deviation value when calculating the deviation value of the operating index.
[0096] By training the local prediction model using historical data within the corresponding geographic area through local servers, the target prediction model output by federated learning is ensured to perform accurate model calculations on the input data within that area, thereby guaranteeing the accuracy of the prediction results.
[0097] S302. Each local server encrypts its own gradient information and uploads it to the central server.
[0098] It should be noted that the central server and local servers communicate via a network. As an example, the central server and multiple local servers can be deployed in a distributed network framework to enable cross-geographical training of the model.
[0099] As an example, the encryption algorithm used by the local server to encrypt the obtained gradient information could be homomorphic encryption.
[0100] As another example, the encryption algorithm used by the local server to encrypt the obtained gradient information can also be a differential privacy algorithm.
[0101] S303. After decrypting the gradient information corresponding to each local server, the central server aggregates the information and determines the global prediction model based on the aggregation result.
[0102] It should be noted that the central server uses the same decryption algorithm as the local servers to decrypt the encrypted data.
[0103] In some implementations of these embodiments, S303 above includes the following steps:
[0104] S3031, The central server aggregates the gradient information of the local prediction models corresponding to multiple geographical regions based on a preset aggregation algorithm to obtain aggregated gradient information; the aggregated gradient information includes aggregated data after aggregating the influence weights of load data, equipment status data and environmental data respectively.
[0105] It should be noted that during the aggregation process, the gradient information corresponding to the load data of different geographical regions is aggregated to obtain aggregated data corresponding to the load data; the gradient information corresponding to the equipment status data of different geographical regions is aggregated to obtain aggregated data corresponding to the equipment status data; and the gradient information corresponding to the environmental data of different geographical regions is aggregated to obtain aggregated data corresponding to the environmental data.
[0106] In some implementations of these embodiments, the central server aggregates the gradient information of local prediction models corresponding to multiple geographical regions based on a preset aggregation algorithm to obtain aggregated gradient information, including:
[0107] First, the central server determines the aggregation weight for each geographical region based on the amount and quality of historical load data, historical equipment status data, and historical environmental data of each power equipment in each geographical region.
[0108] It should be noted that the above data quality can be determined based on data completeness, device type coverage, and historical model contribution.
[0109] The magnitude of the aggregate weight can be used to characterize the relative importance of different geographical regions to operational indicators.
[0110] Second, for each data point in the load data, equipment status data, and environmental data, the weighted average of the impact weights of that data is calculated based on the aggregate weights of each geographical region, thus obtaining the aggregate data corresponding to that data.
[0111] For example, the influence weights corresponding to the load data in the gradient information of each local prediction model can be aggregated using the following formula (2):
[0112] (2);
[0113] In the formula, G represents the aggregated data corresponding to the load data; g i w represents the influence weight corresponding to the load data in the gradient information of the i-th local prediction model. i represents the aggregate weight corresponding to the i-th local prediction model; n represents the number of local prediction models.
[0114] By setting different aggregation weights for different geographical regions, and by aggregating the gradient information of multiple geographical regions into a global prediction model, the output of the global prediction model is guaranteed to be globally optimal, while also ensuring local optimization.
[0115] S3032. Determine the global prediction model based on the aggregated data corresponding to load data, equipment status data, and environmental data.
[0116] S304. In response to the failure to meet the preset convergence condition, update the local prediction models corresponding to each geographic region using the global prediction model, and re-execute the federated learning operation.
[0117] As an example, you can copy the global prediction model and use the copied result as a local prediction model.
[0118] S305. In response to satisfying the preset convergence condition, the global prediction model that satisfies the preset convergence condition is used as the target prediction model.
[0119] As an example, a preset convergence condition could be that the number of iterations of the federated learning operation reaches a preset number of iterations.
[0120] As another example, the preset convergence condition can also be that the aggregated data corresponding to the load data, equipment status data, and environmental data reach a convergence state.
[0121] In these implementations, training the target prediction model in a distributed manner using a federated learning framework can effectively avoid the privacy risks associated with centralized storage of large amounts of data. At the same time, local model training can make full use of the local data attributes of each geographic region to train the model, thereby improving the prediction accuracy of the model.
[0122] Figure 4 This is a schematic diagram of the structure of the power distribution network dynamic reconfiguration device provided in the embodiments of this application, as shown below. Figure 4 As shown, the power distribution network dynamic reconfiguration device 40 provided in this embodiment includes:
[0123] The acquisition module 401 is used to acquire load data, equipment status data and environmental data of multiple substations in the power distribution network; wherein the multiple substations are located in different geographical areas;
[0124] The prediction module 402 is used to input the load data, equipment status data and environmental data of multiple substation equipment into the pre-trained target prediction model. The target prediction model predicts the deviation values of the preset operating indicators of each substation equipment, thereby obtaining the deviation values of the preset operating indicators of multiple substation equipment. The target prediction model is obtained by federated learning of the historical load data, historical equipment status data and historical environmental data of multiple substation equipment in different geographical areas and the historical deviation data of the preset operating indicators.
[0125] The update module 403 is used to update the topology of the distribution network based on the deviation values of the preset operating indicators of multiple power equipment.
[0126] The reconfiguration module 404 is used to reconfigure the distribution network according to the updated topology to obtain the reconfigured distribution network.
[0127] In one possible implementation, the update module 403 is further configured to input the deviation values of the preset operating indicators of the multiple power equipment and the topology of the distribution network into the decision model, and the decision model outputs the updated topology of the distribution network.
[0128] In one possible implementation, the update module 403 is further configured to determine load adjustment data for multiple power equipment based on the deviation value;
[0129] Based on the load adjustment data of multiple power equipment, determine the load adjustment data of power equipment spanning geographical regions;
[0130] Based on load adjustment data of cross-geographical substations and to minimize the total active power loss of the distribution network, the updated distribution network topology is determined.
[0131] In one possible implementation, the distribution network dynamic reconfiguration device 40 further includes a model training module, which is used to learn the gradient information of the local prediction model corresponding to each geographical region based on the historical data of the substation equipment in the geographical region by the local server corresponding to the geographical region.
[0132] Each local server encrypts its own gradient information and uploads it to the central server;
[0133] The central server decrypts the gradient information corresponding to each local server, aggregates it, and determines the global prediction model based on the aggregation result.
[0134] In response to the failure to meet the preset convergence condition, the local prediction models corresponding to each geographic region are updated using the global prediction model, and the federated learning operation is re-executed.
[0135] In response to the satisfaction of the preset convergence condition, the global prediction model that satisfies the preset convergence condition is used as the target prediction model.
[0136] In one possible implementation, the model training module is further configured to, for each geographical region, train a local prediction model for that geographical region using the local server corresponding to that geographical region based on the historical load data, historical equipment status data, historical environmental data, and historical deviation data of preset operating indicators of each power equipment in that geographical region; and obtain the gradient information of the local prediction model in that geographical region.
[0137] The gradient information of the local prediction model includes the influence weights of load data, equipment status data, and environmental data on the deviation values of preset operating indicators.
[0138] In one possible implementation, the model training module is further used by the central server to aggregate the gradient information of the local prediction models corresponding to multiple geographical regions based on a preset aggregation algorithm to obtain aggregated gradient information; the aggregated gradient information includes aggregated data after aggregating the influence weights corresponding to load data, equipment status data and environmental data respectively.
[0139] A global prediction model is determined based on the aggregated data corresponding to load data, equipment status data, and environmental data.
[0140] In one possible implementation, the model training module is also used by the central server to determine the aggregation weights corresponding to each geographical region based on the amount and quality of historical load data, historical equipment status data, and historical environmental data of each power equipment in each geographical region.
[0141] For each data point in the load data, equipment status data, and environmental data, the weighted average of the impact weights of that data is calculated based on the aggregation weights of each geographical region, thus obtaining the corresponding aggregated data.
[0142] In one possible implementation, the operating performance data of the distribution network includes at least one of the following: line loss and voltage.
[0143] In one possible implementation, the update module 403 is further configured to determine the first state of multiple switches in the topology of the distribution network, wherein each switch controls the on / off state of the power transmission line between two substations in the topology of the distribution network.
[0144] The topology of the distribution network with multiple switches in the first state is input into a preset reinforcement learning model, and the reinforcement learning model outputs the second state of the multiple switches; wherein, the second state of the multiple second switches minimizes the total active power loss of the distribution network.
[0145] Update the first state of multiple switches to their respective second states.
[0146] The power distribution network dynamic reconfiguration device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.
[0147] Figure 5 This is a structural schematic diagram of the power distribution network dynamic reconfiguration equipment provided in this application. Figure 5 As shown, the electronic device 50 provided in this embodiment includes at least one processor 501 and a memory 502. Optionally, the device 50 further includes a communication component 503. The processor 501, memory 502, and communication component 503 are connected via a bus.
[0148] In a specific implementation, at least one processor 501 executes computer execution instructions stored in memory 502, causing at least one processor 501 to perform the above-described method.
[0149] The specific implementation process of processor 501 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.
[0150] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.
[0151] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.
[0152] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.
[0153] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.
[0154] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.
[0155] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.
[0156] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.
[0157] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.
[0158] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0159] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0160] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0161] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.
[0162] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.
Claims
1. A method for dynamic reconfiguration of a distribution network, characterized in that, include: The system acquires load data, equipment status data, and environmental data for each of multiple power distribution devices in the power distribution network, wherein the multiple power distribution devices are located in different geographical areas. The load data, equipment status data, and environmental data of each of the multiple substation devices are input into a pre-trained target prediction model. The target prediction model then predicts the deviation values of the preset operating indicators of each of the multiple substation devices, thereby obtaining the deviation values of the preset operating indicators of each of the multiple substation devices. The target prediction model is obtained through federated learning of the historical load data, historical equipment status data, historical environmental data of multiple substation devices in different geographical areas, and the historical deviation data of the preset operating indicators. The topology of the power distribution network is updated based on the deviation values of the preset operating indicators of the multiple power equipment. The distribution network is reconstructed based on the updated topology to obtain the reconstructed distribution network.
2. The method according to claim 1, characterized in that, The step of updating the topology of the distribution network based on the deviation values of the preset operating indicators of the multiple power equipment includes: The deviation values of the preset operating indicators of each of the multiple power equipment and the topology of the distribution network are input into the decision model, and the decision model outputs the updated topology of the distribution network.
3. The method according to claim 2, characterized in that, The updated distribution network topology is determined by the decision model based on the following steps: Based on the deviation value, load adjustment data for multiple power equipment are determined; Based on the load adjustment data of multiple power equipment, determine the load adjustment data of power equipment spanning geographical regions; Based on the load adjustment data of the cross-geographical substations and to minimize the total active power loss of the distribution network, the updated distribution network topology is determined.
4. The method according to any one of claims 1-3, characterized in that, The target prediction model is obtained by performing the following federated learning operation on the local servers and central servers corresponding to each geographical region: For each geographic region, the local server corresponding to that geographic region learns the gradient information of the local prediction model corresponding to that geographic region based on the historical data of the power equipment in that geographic region. Each of the local servers encrypts its own gradient information and uploads it to the central server; The central server decrypts the gradient information corresponding to each of the local servers, aggregates it, and determines the global prediction model based on the aggregation result. In response to the failure to meet the preset convergence condition, the local prediction models corresponding to each geographical region are updated using the global prediction model, and the federated learning operation is re-executed. In response to satisfying the preset convergence condition, the global prediction model that satisfies the preset convergence condition is taken as the target prediction model.
5. The method according to claim 4, characterized in that, For each geographical region, the local server corresponding to that geographical region learns the gradient information of the local prediction model corresponding to that geographical region based on historical data of the power equipment in that geographical region, including: For each geographical region, the local server corresponding to that geographical region trains a local prediction model for that geographical region based on the historical load data, historical equipment status data, historical environmental data, and historical deviation data of preset operating indicators of each power equipment in that geographical region; and obtains the gradient information of the local prediction model for that geographical region. The gradient information of the local prediction model includes the influence weights of load data, equipment status data, and environmental data on the deviation values of preset operating indicators.
6. The method according to claim 5, characterized in that, The central server decrypts the gradient information corresponding to each of the local servers, aggregates it, and determines the global prediction model based on the aggregation result, including: The central server aggregates the gradient information of the local prediction models corresponding to multiple geographical regions based on a preset aggregation algorithm to obtain aggregated gradient information; the aggregated gradient information includes aggregated data after aggregating the influence weights corresponding to the load data, the equipment status data, and the environmental data; The global prediction model is determined based on the aggregated data corresponding to the load data, the equipment status data, and the environmental data.
7. The method according to claim 6, characterized in that, The central server aggregates the gradient information of the local prediction models corresponding to multiple geographical regions based on a preset aggregation algorithm to obtain aggregated gradient information, including: The central server determines the aggregation weight for each geographical region based on the amount and quality of historical load data, historical equipment status data, and historical environmental data of each power equipment in each geographical region. For each data point in the load data, equipment status data, and environmental data, the weighted average of the influence weights of that data is calculated based on the aggregation weights of each geographical region, thus obtaining the aggregated data corresponding to that data.
8. The method according to claim 1, characterized in that, The operational performance data of the power distribution network includes at least one of the following: line loss and voltage.
9. The method according to claim 3, characterized in that, The step of determining the updated distribution network topology based on the load adjustment data of the cross-geographical area substations and minimizing the total active power loss of the distribution network includes: Determine the first state of multiple switches in the topology of the power distribution network, wherein each switch controls the on / off state of the power transmission line between two substations in the topology of the power distribution network. The topology of the distribution network having a first state of multiple switches is input into a preset reinforcement learning model, and the reinforcement learning model outputs a second state of multiple switches; wherein the second state of the multiple second switches minimizes the total active power loss of the distribution network. Update the first state of multiple switches to their respective second states.
10. A dynamic reconfiguration device for a power distribution network, characterized in that, include: The acquisition module is used to acquire load data, equipment status data, and environmental data of multiple substations in the power distribution network; wherein the multiple substations are located in different geographical areas. The prediction module is used to input the load data, equipment status data, and environmental data of each of the multiple substation devices into a pre-trained target prediction model. The target prediction model then predicts the deviation values of the preset operating indicators of each of the multiple substation devices to obtain the deviation values of the preset operating indicators of each of the multiple substation devices. The target prediction model is obtained by federated learning from the historical load data, historical equipment status data, historical environmental data of multiple substation devices in different geographical areas, as well as the historical deviation data of the preset operating indicators. The update module is used to update the topology of the power distribution network based on the deviation values of the preset operating indicators of the multiple power equipment. The reconfiguration module is used to reconfigure the distribution network according to the updated topology to obtain the reconfigured distribution network.
11. A dynamic reconfiguration device for a power distribution network, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-9.
12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-9.
13. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method described in any one of claims 1-9.