Power distribution network reconstruction method and system adaptive to source network load storage coordination

Abstract

The invention relates to a power distribution network reconstruction method and system adaptive to source network load storage coordination, and the method comprises the steps: obtaining a network structure of a target power distribution network, and determining all network reconstruction schemes combined by switching states; constructing a scoring system for the network reconstruction scheme based on the state indexes of the power distribution network; power distribution network power distribution data is collected, a power distribution network state index is calculated, and then a scoring result of a network reconstruction scheme is obtained through a scoring system; based on each network reconstruction scheme, each corresponding power distribution data and the scoring result as input data, training through a pattern recognition algorithm to obtain mapping of the network reconstruction scheme and the power distribution data thereof to the scoring result; and taking actually measured or predicted power distribution data as input, scoring all the network reconstruction schemes through a pattern recognition algorithm, and selecting the network reconstruction scheme with the highest score for reconstruction. Compared with the prior art, the problems of heavy calculation burden, overfitting, sensitivity to initial conditions and the like are avoided.

Description

Technical Field

[0001] This invention relates to the field of distribution network reconfiguration technology, and in particular to a distribution network reconfiguration method and system adapted to source-grid-load-storage coordination. Background Technology

[0002] The increasing integration of renewable energy has brought a series of problems to distribution network systems, including voltage exceeding limits, line congestion, and transformer overload. Besides introducing a source-grid-load-storage collaborative model, adopting appropriate grid management strategies is also crucial for improving grid flexibility and resilience. Specifically, reconfiguring the distribution network through switching can reduce power losses and equipment overload, improve voltage distribution, balance related loads, and promote reliable and economical operation of the distribution network. Therefore, advanced tools and methods can be used to determine the optimal topology of the distribution network, and automation can be achieved through real-time operation and control methods.

[0003] However, even for medium-sized networks, using traditional methods to enumerate all possible configurations and select the optimal result under multiple objectives leads to an exponential increase in computational cost.

[0004] In addition, some methods transform multi-objective optimization problems into single-objective optimization problems, decomposing the reconfiguration problem into a complex combinatorial problem due to current and voltage constraints. For example, the invention disclosed in CN117154726A, a distribution network reconfiguration and power restoration method and system based on island partitioning, suffers from limitations such as high data requirements and poor optimization results. Therefore, a new distribution network reconfiguration method adapted to source-grid-load-storage coordination is needed to avoid problems such as heavy computational burden, overfitting, and sensitivity to initial conditions. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a distribution network reconfiguration method and system that adapts to the coordination of source, grid, load and storage, avoiding problems such as heavy computational burden, overfitting and sensitivity to initial conditions.

[0006] The objective of this invention can be achieved through the following technical solutions: A distribution network reconfiguration method adapted to source-grid-load-storage coordination includes the following steps: Obtain the network structure of the target distribution network and determine all network reconfiguration schemes based on the combination of switch states; construct a scoring system for the network reconfiguration schemes based on predefined distribution network state indicators; Power distribution data of the distribution network is collected for each network reconfiguration scheme. The voltage distribution of each node in the distribution network is obtained through power flow calculation to calculate the distribution network status index. Then, the scoring result of the network reconfiguration scheme is obtained through the scoring system. Based on each network reconstruction scheme, the corresponding power distribution data and scoring results, the data are used as input data to train a pattern recognition algorithm to obtain the mapping from the network reconstruction scheme and its power distribution data to the scoring results. Using measured or predicted power distribution data of the target distribution network as input, a trained pattern recognition algorithm scores all possible network reconfiguration schemes and selects the network reconfiguration scheme with the highest score for reconfiguration.

[0007] Furthermore, the distribution network status index is the over-limit status of nodes; the evaluation process for the over-limit status of nodes is as follows: The voltage distribution results of each node calculated based on the power distribution data of the distribution network are compared with the preset upper and lower limits of the operating voltage to determine whether each node has an overvoltage or undervoltage condition.

[0008] Furthermore, the method divides all nodes in the distribution network into A cluster of nodes with similar topological locations and voltage characteristics is used to determine the over-limit status on a cluster basis.

[0009] Furthermore, for a number of switches... The distribution network is obtained by combining switch states. Network reconstruction schemes; The number of scoring results is: In the formula, The number of scoring results.

[0010] Furthermore, the scoring rules of the scoring system are as follows: When N clusters exhibit out-of-limit violations, various network reconstruction schemes are traversed. As the number of closed switches increases, the score decreases progressively. .

[0011] Furthermore, the scoring system is used to obtain all possible scores for all network reconstruction schemes under all possible conditions of exceeding limits in the cluster, in order to construct an output dataset for training the pattern recognition algorithm.

[0012] Furthermore, the output dataset includes multiple feature vectors, each of which includes power distribution data of the distribution network, the switching status of the corresponding network reconfiguration scheme, and a score value.

[0013] Furthermore, the pattern recognition algorithm employs an ensemble learning model for prediction. This ensemble learning model uses multiple base models to achieve combined prediction, and the corresponding expression is: In the formula, These are the predicted values ​​from the ensemble learning model. For the first A basic model It is the total number of basic models. It is a model definition The parameters, The input feature vector, This is the predicted result.

[0014] Furthermore, the loss function of the ensemble learning model in the prediction process is: In the formula, The loss function of the ensemble learning model. The loss between the predicted values ​​and the corresponding true predictions of the ensemble learning model. Indicates the relationship between input and output. To the target Regularization terms related to the mapping function, , , and These are the parameters that define the model.

[0015] The present invention also provides a distribution network reconfiguration system adapted to source-grid-load-storage coordination, including a memory and a processor. The memory stores a computer program, and the processor calls the computer program to execute the steps of the method described above.

[0016] Compared with the prior art, the present invention has the following advantages: (1) This invention first determines all possible types of network reconfiguration schemes based on the network structure of the target distribution network, and constructs a scoring system for distribution network reconfiguration schemes based on distribution network status indicators. A database is built based on all different network reconfiguration schemes, distribution network status indicators, and the scores calculated by the scoring system, for training based on a pattern recognition algorithm. The training results are used to score various distribution network reconfiguration schemes based on measured or predicted distribution network operation data, and the optimal distribution network reconfiguration scheme is obtained by comparing the scoring results. By applying this method, this scheme avoids problems such as heavy computational burden, overfitting, and sensitivity to initial conditions, providing an innovative solution to the distribution network reconfiguration problem and offering important guidance for distribution network planning. The effectiveness of the model is verified through implementation examples, clarifying the characteristics and advantages of this scheme.

[0017] (2) The present invention uses an integrated learning model to score and predict the distribution network reconfiguration scheme. The integrated learning model uses multiple basic models to achieve combined prediction, and the training effect on the dataset constructed from multiple distribution network reconfiguration schemes, distribution network status and scoring results is significant. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a distribution network reconfiguration method adapted to source-grid-load-storage coordination provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of a power distribution network structure provided in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0022] Example 1 like Figure 1 As shown, this embodiment provides a distribution network reconfiguration method adapted to source-grid-load-storage coordination, including the following steps: S1: Obtain the network structure of the target distribution network and determine all network reconfiguration schemes based on the combination of switch states; construct a scoring system for the network reconfiguration schemes based on predefined distribution network state indicators; S2: Collect power distribution data of the distribution network for each network reconfiguration scheme, obtain the voltage distribution of each node in the distribution network through power flow calculation, calculate the distribution network status index, and then obtain the scoring results of the network reconfiguration scheme through the scoring system. S3: Based on each network reconstruction scheme, the corresponding power distribution data and scoring results, as input data, a pattern recognition algorithm is used for training to obtain the mapping from the network reconstruction scheme and its power distribution data to the scoring results; S4: Using measured or predicted power distribution data of the target distribution network as input, a trained pattern recognition algorithm scores all possible network reconfiguration schemes and selects the network reconfiguration scheme with the highest score for reconfiguration.

[0023] In step S1, the process of obtaining the network structure of the target distribution network can be as follows: For the main transformer substation and its downstream medium-voltage distribution network, the number of distribution network nodes, lines, and switches are defined as follows: , and The maximum number of combinations of switch states can be obtained. A network reconfiguration scheme is proposed. Based on the access status of distributed resources (energy generation, load, and storage) in the distribution network, the number of distributed energy storage, photovoltaic power generation, wind power generation, and loads connected are defined as follows: , , and .

[0024] The distribution network status index can be the over-limit status of nodes; the evaluation process for the over-limit status of nodes is as follows: The voltage distribution results of each node calculated based on the power distribution data of the distribution network are compared with the preset upper and lower limits of the operating voltage to determine whether each node has an overvoltage or undervoltage condition.

[0025] For example, by acquiring active and reactive power measurement data from each device node, and setting the high-voltage side of the main transformer as a balancing node, the voltage distribution of the distribution network can be calculated, and then based on preset upper and lower limits of the operating voltage... and It determines whether each node is in an overvoltage or undervoltage state.

[0026] Preferably, for the purpose of improving algorithm efficiency and reducing information dimensionality, cluster analysis is used to divide all nodes into... A cluster of nodes with similar topological locations and voltage characteristics is used to determine the voltage over-limit status on a cluster basis.

[0027] Based on distribution network status indicators, a distribution network reconfiguration scheme-operation status scoring system is constructed, providing an operation status scoring matrix for different distribution network reconfiguration schemes and their node clusters. The number of rows in the matrix represents the number of network reconfiguration schemes. The pure radial configuration is located in the first row, followed by configurations arranged in ascending order of the number of closed switches in each scheme: radial, scheme 1, ..., scheme The number of columns in the matrix represents the number of node clusters. Arranged in order of severity of voltage violations: This indicates a voltage failure; This indicates that one cluster has experienced a voltage violation; ...; until This indicates that voltage violations exist in all clusters.

[0028] In the above rating matrix, the number of rating results is: In the formula, The number of scoring results.

[0029] The scoring rules of the scoring system are as follows: When N clusters exhibit out-of-limit violations, various network reconstruction schemes are traversed. As the number of closed switches increases, the score decreases progressively. .

[0030] Specifically, in the scoring system, the highest score This corresponds to the optimal operating state, i.e., no voltage violations under the radial configuration. The score decreases sequentially according to the following rules: First, maintaining no voltage violations ( Under the premise of (state), as the number of closed switches increases, the score decreases progressively; After all state reconstruction schemes are completed, the scoring sequence is transferred to... The status is then determined, and the score decreases sequentially from the radial configuration, with subsequent statuses following the same pattern until all combinations of refactoring schemes and running states are covered, with the lowest score being 1.

[0031] In step S2, the main purpose is to collect power distribution data of the distribution network, obtain the voltage distribution of each node in the distribution network through power flow calculation, calculate the distribution network status index, and output the final network configuration score.

[0032] The scoring system is used to obtain all possible scores for all network reconstruction schemes under all possible conditions of exceeding limits in the cluster, in order to construct an output dataset for training the pattern recognition algorithm.

[0033] The output dataset includes multiple feature vectors, each of which includes power distribution data of the distribution network, the switching status of the corresponding network reconfiguration scheme, and a score value.

[0034] The specific implementation process includes: constructing a power distribution dataset; based on this dataset, obtaining the voltage distribution of each node in the distribution network through power flow calculation; then, according to the scoring system constructed in step S1, obtaining the score value of each power grid reconfiguration scheme and its corresponding power grid operating state; merging the power distribution dataset and the score values ​​to generate an output dataset, i.e., a feature vector. The feature vector contains the following elements: 1) All Power distribution data for each grid-connected device, including: The active power output of photovoltaic power generation Active power output of wind power generation The active power consumption at the load, and The active power of energy storage charging and discharging; 2) All in the distribution network The state variables of each switch are represented by binary discriminant variables; 3) The scoring value obtained by the scoring system for all possible combinations of switch state combinations and cluster voltage violation states.

[0035] The constructed feature vectors can be used for subsequent scoring algorithm training, where power distribution data and switching state data are the input data for algorithm training, and the scoring result is the target of algorithm training.

[0036] In step S3, optionally, the pattern recognition algorithm uses an ensemble learning model for prediction. The ensemble learning model uses multiple basic models to achieve combined prediction and constructs a corresponding loss function.

[0037] The overall process includes: Using the reconfiguration scheme, operating status, and corresponding scoring results as input data, and a portion of the samples as the training set, a scoring algorithm is trained based on a pattern recognition algorithm to establish a mapping from the distribution network reconfiguration scheme and its operating status data to the scoring results; the remaining samples are used as the test set to verify the algorithm's training effect.

[0038] Specifically, the feature vectors described in step S2 are used as input to the training algorithm, and a portion of the samples are divided into a training set for the training phase of the algorithm, while the other portion is divided into a test set for the validation of the algorithm. Training is implemented through a pattern recognition algorithm, which includes an ensemble learning method to achieve combined prediction through a series of relatively simple basic models. Specifically, for a given dataset... ,Depend on Composed of 1 sample, let For feature vectors, Let be the actual value of the target, corresponding to the l-th sample. The model's task is to learn from the input (feature vector). From output (target) Mapping function of ) The ensemble learning model takes the following form: In the formula, These are the predicted values ​​from the ensemble learning model. For the first A basic model It is the total number of basic models. It is a model definition The parameters, The input feature vector, This is the predicted result.

[0039] Then, by setting a loss function to minimize the empirical loss and avoid the risk of overfitting, the empirical loss function is minimized to the true value. Compared with the predicted value The difference between them is expressed by the following formula: In the formula, The loss function of the ensemble learning model. The loss between the predicted values ​​and the corresponding true predictions of the ensemble learning model. Indicates the relationship between input and output. To the target Regularization terms related to the mapping function, , , and These are the parameters that define the model.

[0040] For the model The loss is given by the following formula: In this embodiment, the specific implementation method of the above process is as follows: The specific implementation is based on the IEEE 33-node distribution network, and its structure is as follows: Figure 2 As shown, the network contains 33 nodes ( ) and 32 lines ( These lines can be connected by 3 switches ( The control of the three switches is integrated into the radial distribution network, and the three switches are named according to the nodes they are connected to. , and Enumerating all combinations of switch states, the network has a total of Eight possible configurations. Using 1 / 0 to represent switch open / closed, the correspondence between these eight power grid configurations is shown in Table 1.

[0041] Table 1 Distribution Network Reconfiguration Scheme The nodes are divided into 4 clusters, namely , , , Each node in the network is connected to one of the following devices: photovoltaic (PV), wind turbine (WT), load, or distributed energy storage (ESS). For all power flow calculations, node 1 is defined as the slack node, and the remaining nodes are considered PQ nodes, with their active and reactive power values ​​being the input data from the dataset.

[0042] Then, power flow calculations are performed on the dataset composed of historical data to obtain the voltage distribution of each node in the distribution network. If one node in the cluster experiences overvoltage, the value for that cluster is set to +1; if one node experiences undervoltage, it is set to -1; if the voltage of all nodes is within the range of 0.95 pu-1.05 pu (representing respectively...). and If the value is 0, then it is set to 0. With 4 clusters, 5 cluster states will be exhibited: 0, 1, 2, 3, and 4 clusters exceeding voltage limits, resulting in 40 combinations of switching and cluster states. According to the formula... It is concluded that, under optimal conditions, Thus, the scoring matrix is ​​obtained, as shown in Table 2.

[0043] Table 2 Reconstruction Scheme Scoring Matrix After assigning scores to each cluster state and switch configuration, the power curve dataset is merged with the assigned scores to generate the output dataset. 75% of the samples are used to train the algorithm, and the remaining 25% are used to test it.

[0044] Step S4 specifically involves: using the predicted or measured power data of each node in the distribution network as input, with the format of the input data consistent with the format of the power distribution dataset; performing score prediction on all possible network configurations based on the scoring algorithm proposed in step S3; and determining the network configuration with the highest score as the optimal network reconfiguration scheme under the current operating conditions based on the score prediction results.

[0045] The specific implementation process of this embodiment is as follows: The measured power data of each node in the distribution network is used as input, as shown in Table 3. The format of the input data is consistent with that of the power distribution dataset. Based on the scoring algorithm proposed in step S3, all possible network configurations are scored and predicted to obtain the scores corresponding to different reconfiguration schemes, as shown in Table 4. According to the scoring prediction results, reconfiguration scheme 4 has the highest score, which is determined to be the optimal network reconfiguration scheme under the current operating conditions.

[0046] Table 3 Power of Distribution Network Nodes Table 4 Restructuring Scheme Scoring Example 2 This embodiment provides a distribution network reconfiguration system adapted to source-grid-load-storage coordination, including a memory and a processor. The memory stores a computer program, and the processor calls the computer program to execute the steps of a distribution network reconfiguration method adapted to source-grid-load-storage coordination as described in Embodiment 1.

[0047] The computer program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This computer program code can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the computer program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0048] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A distribution network reconfiguration method adapted to source-grid-load-storage coordination, characterized in that, Includes the following steps: Obtain the network structure of the target distribution network and determine all network reconfiguration schemes based on the combination of switch states; construct a scoring system for the network reconfiguration schemes based on predefined distribution network state indicators; Power distribution data of the distribution network is collected for each network reconfiguration scheme. The voltage distribution of each node in the distribution network is obtained through power flow calculation to calculate the distribution network status index. Then, the scoring result of the network reconfiguration scheme is obtained through the scoring system. Based on each network reconstruction scheme, the corresponding power distribution data and scoring results, as input data, a pattern recognition algorithm is used to train and obtain the mapping from the network reconstruction scheme and its power distribution data to the scoring results. Using measured or predicted power distribution data of the target distribution network as input, a trained pattern recognition algorithm scores all possible network reconfiguration schemes and selects the network reconfiguration scheme with the highest score for reconfiguration.

2. The distribution network reconfiguration method adapting to source-grid-load-storage coordination according to claim 1, characterized in that, The distribution network status index is the over-limit status of the nodes; the evaluation process for the over-limit status of the nodes is as follows: The voltage distribution results of each node calculated based on the power distribution data of the distribution network are compared with the preset upper and lower limits of the operating voltage to determine whether each node has an overvoltage or undervoltage condition.

3. The distribution network reconfiguration method adapting to source-grid-load-storage coordination according to claim 2, characterized in that, The method divides all nodes in the distribution network into A cluster of nodes with similar topological locations and voltage characteristics is used to determine the over-limit status on a cluster basis.

4. The distribution network reconfiguration method adapting to source-grid-load-storage coordination according to claim 3, characterized in that, For the number of switches is The distribution network is obtained by combining switch states. Network reconstruction schemes; The number of scoring results is: In the formula, The number of scoring results.

5. A distribution network reconfiguration method adapting to source-grid-load-storage coordination according to claim 3, characterized in that, The scoring rules of the scoring system are as follows: When N clusters exhibit out-of-limit violations, various network reconstruction schemes are traversed. As the number of closed switches increases, the score decreases progressively. .

6. A distribution network reconfiguration method adapting to source-grid-load-storage coordination according to claim 5, characterized in that, The scoring system is used to obtain all possible scores for all network reconstruction schemes under all possible conditions of exceeding limits in the cluster, in order to construct an output dataset for training the pattern recognition algorithm.

7. A distribution network reconfiguration method adapting to source-grid-load-storage coordination according to claim 6, characterized in that, The output dataset includes multiple feature vectors, each of which includes power distribution data of the distribution network, the switching status of the corresponding network reconfiguration scheme, and a score value.

8. A distribution network reconfiguration method adapting to source-grid-load-storage coordination according to claim 1, characterized in that, The pattern recognition algorithm uses an ensemble learning model for prediction. This ensemble learning model employs multiple base models to achieve combined prediction, and the corresponding expression is: In the formula, These are the predicted values ​​from the ensemble learning model. For the first A basic model It is the total number of basic models. It is a model definition The parameters, The input feature vector, This is the predicted result.

9. A distribution network reconfiguration method adapting to source-grid-load-storage coordination according to claim 8, characterized in that, The loss function of the ensemble learning model in the prediction process is: In the formula, The loss function of the ensemble learning model. The loss between the predicted values ​​and the corresponding true predictions of the ensemble learning model. Indicates the relationship between input and output. To the target Regularization terms related to the mapping function, , , and These are the parameters that define the model.

10. A distribution network reconfiguration system adapted to source-grid-load-storage coordination, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and the processor invokes the computer program to perform the steps of the method as described in any one of claims 1 to 9.

Images