Power distribution network reconstruction method and system considering dynamic line loss treatment

Abstract

The invention discloses a power distribution network reconstruction method considering dynamic line loss treatment, and the method comprises the steps: S1, obtaining the data of a power distribution network, and carrying out the preprocessing of the data; s2, constructing a line power flow distribution prediction model based on deep learning, carrying out training based on the preprocessed power distribution network data, and carrying out power flow distribution calculation by using the trained line power flow distribution prediction model to output a line power flow distribution prediction result; s3, constructing a dynamic line loss evaluation model of the power distribution network, so as to obtain the dynamic line loss of the power distribution network in each flow direction based on the power flow distribution prediction result; s4, taking the minimum treatment equivalent cost of the dynamic line loss as a target, considering a plurality of constraint conditions, and constructing a network reconstruction optimization model of the power distribution network; and S5, solving the power distribution network reconstruction optimization model to obtain all power distribution network reconstruction strategies meeting all constraint conditions, and selecting and outputting an optimal power distribution network reconstruction strategy.

Description

Technical Field

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

[0002] In recent years, in the face of the challenges brought by the growth of power load and the complexity of the power grid, distribution network reconstruction technology, as one of the important means to optimize the operation of the distribution system, can effectively balance the load, eliminate overload, and quickly restore power supply in the event of a fault, thereby greatly improving the overall performance of the distribution system; among them, distribution network reconstruction can change the topology of the distribution network by switching the state of the connecting switch and the section switch, thereby affecting the distribution direction of the power flow in the distribution network, so as to reduce network losses and improve the power supply quality and system reliability.

[0003] In the current existing technologies, research on distribution network reconstruction mainly focuses on the following aspects: 1. Research on distribution network reconstruction considering the access of distributed power sources; The traditional distribution network structure and operation control method are difficult to adapt to the access of a large number of distributed power sources with uncertain characteristics. Based on the impact of the access capacity, location and power factor of distributed power sources on network losses, the distribution network reconstruction optimization method of distribution network flow calculation containing distributed power sources is studied.

[0004] 2. Research on distribution network reconstruction considering demand response; Based on the impact of time-of-use electricity prices on user loads, the ability of electric vehicle charging loads, user-side energy storage, large industrial adjustable loads, etc. to participate in demand response is comprehensively considered to construct a distribution network reconstruction optimization method with the goal of minimizing active network losses.

[0005] 3. Research on distribution network fault recovery and reconstruction methods: With the aim of improving the rapid recovery capability and power supply reliability of the distribution network after a fault occurs, a distribution network reconstruction method is proposed in combination with the real-time operating status and topological structure of the distribution network, and the power supply recovery strategy in the fault area is studied.

[0006] The study found that some scholars have conducted research on distribution network reconstruction methods for distributed power sources and new load access, but there is still room for improvement: First, traditional distribution network reconstruction methods are mainly static network reconstruction, but in the actual operation of the distribution network, the loads of each branch are constantly changing, and are limited by the number of opening and closing of section switches and connecting switches, making it difficult to achieve accurate dynamic network reconstruction; second, when constructing the reconstruction optimization method, typical economic indicators such as the minimum number of switch operations and the minimum network operation loss are often used as optimization goals, which does not conform to the development of the power market and take the adjustable resources on the user side into consideration, making it difficult to reflect the objective optimization needs.

[0007] Therefore, there is an urgent need to explore a new distribution network reconstruction method and system that can not only meet the dynamic line loss management needs of the distribution network under the condition of random load changes on each branch at each moment, but also ensure the reliable operation of the distribution system and minimize the objective equivalent cost of network reconstruction. Summary of the invention

[0008] The technical problem to be solved by the present invention is: the present invention discloses a distribution network network reconstruction method and system taking dynamic line loss management into consideration, so as to solve the problem that the existing distribution network network reconstruction method cannot both meet the dynamic line loss management needs of the distribution network under the condition of random load changes of each branch at each moment and ensure the reliable operation of the distribution system and minimize the equivalent cost of network reconstruction.

[0009] In order to solve the above technical problems, the technical solution adopted by the present invention is: a distribution network reconstruction method considering dynamic line loss management, which comprises the steps of: S1: Obtain distribution network data and perform preprocessing; S2: Construct a line flow distribution prediction model based on deep learning, train it based on the preprocessed distribution network data, and then use the trained line flow distribution prediction model to carry out flow distribution calculation to output the line flow distribution prediction result; S3: constructing a distribution network dynamic line loss assessment model to obtain the dynamic line loss of each flow direction of the distribution network flow distribution based on the flow distribution prediction result; S4: Taking the minimization of the equivalent cost of dynamic line loss management as the goal, considering multiple constraints, a distribution network reconstruction optimization model is constructed; S5: Solve the distribution network reconstruction optimization model to obtain all distribution network reconstruction strategies that meet all constraints, select and output the optimal distribution network reconstruction strategy.

[0010] In the present invention, the inventors have optimized and designed a new distribution network reconstruction method that takes into account dynamic line loss management. The distribution network reconstruction method can comprehensively consider key factors such as electric vehicle load changes and distributed power supply access. A deep learning optimization algorithm can be used to construct a line flow distribution prediction model based on deep learning, thereby realizing real-time prediction of the flow of each branch.

[0011] At the same time, in the present invention, a dynamic line loss assessment model for the distribution network is also constructed to accurately predict the dynamic network losses in each flow direction of the distribution network flow distribution; in addition, in the present invention, it takes the minimization of the equivalent cost of dynamic line loss management as the goal, considers the user-side energy storage call cost and the distribution network reconstruction cost in the region, and takes into account the power supply reliability and network flexibility of the distribution network, constructs a distribution network reconstruction optimization model, and finally can formulate the optimal distribution network reconstruction strategy that meets all constraints.

[0012] To sum up, the distribution network reconstruction method considering dynamic line loss management described in the present invention can significantly improve the operating efficiency of the distribution network, reduce power loss and operating costs, promote energy conservation and emission reduction and technological innovation in the power industry, and has good promotion prospects and application value.

[0013] Furthermore, in the distribution network reconstruction method of the present invention, in step S1, the distribution network data includes: Distribution network structure, branch parameters, node data, user historical load data, and temperature, humidity, working day and holiday data consistent with the time scale of user historical load data.

[0014] Furthermore, in the distribution network reconstruction method of the present invention, in step S1, the preprocessing includes: A local outlier factor detection algorithm is used to identify abnormal load data and missing value data in the user's historical load data, and a sliding average method is used for processing.

[0015] Furthermore, in the distribution network reconstruction method of the present invention, in step S2, the following steps are specifically included: S21: acquiring load sequence data based on the preprocessed distribution network data, normalizing the load sequence data, and dividing the normalized load sequence data into two parts according to the time series, one part as a training set and the other part as a test set; S22: Construct a line flow distribution prediction model based on deep learning; S23: inputting the training set into the line power flow distribution prediction model to perform model training; S24: Inputting the test set into the trained line power distribution prediction model to output the line power distribution prediction result.

[0016] Furthermore, in the distribution network reconstruction method described in the present invention, in step S22, a branch prediction network is established using a GRU network, and all branch prediction networks constitute the line flow distribution prediction model, while sharing the total load characteristics to link all branch prediction networks together.

[0017] Furthermore, in the distribution network reconstruction method of the present invention, step S24 is specifically as follows: The test set is input into the trained line flow distribution prediction model to obtain the load prediction result at time t by denormalization. , and obtain the load forecast result of branch n at time t , to obtain the power flow of branch n at time t ;

[0018] in, is the power flow of branch n at time t; is the voltage of branch n at time t.

[0019] Furthermore, in the distribution network reconstruction method described in the present invention, in step S4, the constraint conditions include: power flow constraint, node voltage constraint, branch capacity constraint, reconstruction number constraint, and energy storage call constraint.

[0020] Furthermore, in the distribution network reconstruction method of the present invention, in step S5, specifically: The MISOCP optimization algorithm is applied to solve the distribution network reconstruction optimization model to obtain all distribution network reconstruction strategies that meet all constraints, and the optimal distribution network reconstruction strategy is selected and output.

[0021] Accordingly, the present invention also discloses a distribution network reconstruction system considering dynamic line loss management, and the distribution network reconstruction system can be specifically used to implement the distribution network reconstruction method of the present invention, which specifically includes: Data input and preprocessing module, which is used to obtain distribution network data and perform preprocessing; The branch power flow analysis and prediction module is used to construct a line power flow distribution prediction model based on deep learning, which is trained based on the preprocessed distribution network data, and then uses the trained line power flow distribution prediction model to carry out power flow distribution calculation to output the line power flow distribution prediction result; A dynamic line loss calculation and evaluation module, which is used to construct a dynamic line loss evaluation model for the distribution network, so as to obtain the dynamic line loss of each flow of the distribution network flow distribution based on the power flow distribution prediction result; The distribution network reconstruction optimization module is used to construct a distribution network reconstruction optimization model with the goal of minimizing the equivalent cost of dynamic line loss management and considering multiple constraints; The decision implementation and monitoring feedback module is used to solve the distribution network reconstruction optimization model to obtain all distribution network reconstruction strategies that meet all constraints, select and output the optimal distribution network reconstruction strategy.

[0022] Furthermore, in the distribution network reconstruction system described in the present invention, the decision implementation and monitoring feedback module can also monitor the operating status of the distribution network in real time, and adjust the distribution network reconstruction strategy in real time according to the feedback.

[0023] The beneficial effects of the present invention are as follows: the present invention discloses a distribution network network reconstruction method and system taking dynamic line loss management into consideration, which can effectively solve the technical problems existing in the existing distribution network network reconstruction method, and can not only meet the dynamic line loss management requirements of the distribution network under the condition of random load changes of each branch at each moment, but also ensure the reliable operation of the distribution system and the minimum equivalent cost of network reconstruction. It can significantly improve the operating efficiency of the distribution network, reduce power loss and operating costs, and promote energy conservation, emission reduction and technological innovation in the power industry. It has good promotion prospects and application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The regional power flow distribution diagram of the distribution network is schematically shown.

[0025] Figure 2 The present invention is a flowchart of the steps of a distribution network reconstruction method considering dynamic line loss management in one implementation mode.

[0026] Figure 3 This is a module design diagram of a distribution network reconstruction system taking dynamic line loss management into consideration according to the present invention in one implementation manner.

[0027] Figure 4 The present invention is a flow chart of the steps of implementing the distribution network reconstruction method described above by the distribution network reconstruction system considering dynamic line loss management. DETAILED DESCRIPTION

[0028] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in conjunction with the implementation modes and the accompanying drawings.

[0029] Please refer to Figure 1 and Figure 2 As shown, in this embodiment, the present invention discloses a distribution network reconstruction method considering dynamic line loss management, which specifically includes the steps of: S1: Obtain distribution network data and perform preprocessing; S2: Construct a line flow distribution prediction model based on deep learning, train it based on the preprocessed distribution network data, and then use the trained line flow distribution prediction model to carry out flow distribution calculation to output the line flow distribution prediction result; S3: constructing a distribution network dynamic line loss assessment model to obtain the dynamic line loss of each flow direction of the distribution network flow distribution based on the flow distribution prediction result; S4: Taking the minimization of the equivalent cost of dynamic line loss management as the goal, considering multiple constraints, a distribution network reconstruction optimization model is constructed; S5: Solve the distribution network reconstruction optimization model to obtain all distribution network reconstruction strategies that meet all constraints, select and output the optimal distribution network reconstruction strategy.

[0030] That is to say, in this embodiment, the distribution network reconstruction method disclosed in the present invention can first perform real-time prediction of the current of each branch in the area of ​​the distribution network through line flow distribution prediction based on deep learning, so as to construct a line flow distribution prediction model based on deep learning, and then evaluate the dynamic network loss. According to the evaluation results, a distribution network dynamic line loss evaluation model is constructed, and with the goal of minimizing the equivalent cost of dynamic line loss management, a distribution network network reconstruction optimization model is constructed considering multiple constraints. By solving the distribution network network reconstruction optimization model, the optimal distribution network reconstruction strategy can be obtained, thereby ensuring the economic operation of the distribution network system while achieving accurate dynamic network reconstruction.

[0031] Combined with reference Figure 1 and Figure 2 In this embodiment, in the distribution network reconstruction method considering dynamic line loss management of the present invention, in the above step S1, specifically: Input the distribution network data in a certain area. The distribution network data can be specifically the historical data of the distribution network. The historical data of the distribution network specifically includes: the network structure of the distribution network, branch parameters and node data. Where G represents the distribution network in the area; G = (V, E) represents the topological structure of the distribution network; V represents the branch set, V = {v1, v2,…, v N}, that is, within the distribution network area G, there are a total of N branches.

[0032] Correspondingly, the input distribution network data also includes the user historical load data, which contains the historical hourly load data of all users in each branch, denoted as P it , P it represents the user load data of the i-th branch at time t. Meanwhile, in this embodiment, the input distribution network data also includes temperature, humidity, working day and holiday data consistent with the time scale of the user's historical load data.

[0033] It should be noted that in the above step S1 of the present invention, when the above distribution network data is input, the distribution network data will also be pre-processed. For example, in this embodiment, it is necessary to specifically adopt a local outlier factor detection algorithm to distinguish abnormal load data and missing value data in the user's historical load data, and use a sliding average method to process it to supplement the incomplete data or abnormal data.

[0034] Accordingly, in the distribution network reconstruction method of the present invention, in the above step S2, a line power flow distribution prediction model based on deep learning can be constructed, and the line power flow distribution prediction result can be output, which can specifically include the following steps S21-S24; S21: acquiring load sequence data based on the preprocessed distribution network data, normalizing the load sequence data, and dividing the normalized load sequence data into two parts according to the time series, one part as a training set and the other part as a test set.

[0035] For example, in this embodiment, the load sequence data can be obtained from the preprocessed distribution network data. To realize the dynamic line loss assessment management of the distribution network, the load of all users on the i-th branch at each moment needs to be predicted. represents the load sequence of the i-th branch. The load state in region G can be expressed as the matrix P K×N ,like Figure 1 As shown, the region G may include K1-K N , K represents the length of each branch load sequence. Suppose the load data sequence after preprocessing is set to P = [P1, P2,… P T ], P t ∈R N×t It represents the sum of the loads of all branches at time t, that is, the characteristic matrix at time t.

[0036] In this embodiment, after the load sequence data is obtained, in step S21, max-min normalization processing can be performed on the load sequence data, and the load sequence data can be divided into two parts according to the time series using a sliding window method, 80% as a training set and 20% as a test set.

[0037] S22: Construct a line flow distribution prediction model based on deep learning.

[0038] In the above step S22 of the present invention, according to the definition of deep learning, the GRU network can be specifically used to establish a branch prediction network, and all branch prediction networks constitute the line flow distribution prediction model, while sharing the total load characteristics to link all branch prediction networks together.

[0039] S23: Inputting the training set into the line power flow distribution prediction model to perform model training.

[0040] In the above step S23 of the present invention, a training set can be selected for training, and the total load data P t And each branch load data P it After fusion, the corresponding branch GRU network is forward trained using the corresponding fusion data until convergence.

[0041] S24: Inputting the test set into the trained line power distribution prediction model to output the line power distribution prediction result.

[0042] In the present invention, in this embodiment, see Figure 1 In the region G shown in FIG. 1 , there are N branches in the region G. In the above step S24, the test set is input into the trained line flow distribution prediction model to obtain the load prediction result at time t by denormalization. , which can be specifically:

[0043] Accordingly, the load forecast result of branch n at time t can be further obtained: , the load forecast result of the nth branch at time t It can be expressed as:

[0044] Hence, see Figure 1 Considering that all loads are connected to the 10kV voltage level, the relationship between the branch input power and voltage can be used to obtain the power flow of branch n at time t ;

[0045] in, is the power flow of branch n at time t; is the voltage of branch n at time t.

[0046] Based on this, in the present invention, Figure 1 The power flow in the distribution network G area shown can be expressed as:

[0047] Therefore, through S21-S24, a line flow distribution prediction model based on deep learning can be constructed first, and the line flow distribution prediction results can be output to make real-time predictions of the flow of each branch in the distribution network area.

[0048] Similarly, in the distribution network reconstruction method, in step S3, it is possible to construct a distribution network dynamic line loss assessment model and obtain the dynamic line loss of each flow direction of the distribution network flow distribution using the flow distribution prediction result output by the line flow distribution prediction model.

[0049] In this embodiment, based on the power flow calculation results, the loss of the nth branch at time t can be calculated: :

[0050] in, represents the loss of the nth branch at time t, Represents the resistance of the nth branch.

[0051] Then, the total line loss in the distribution network area G is It can be expressed as:

[0052] Of course, considering the increased uncertainty of power flow caused by load changes and the changes in distribution network line losses caused by local unequal operation, the load distribution can be adjusted by adjusting the line interconnection switch or calling the user-side energy storage, thereby optimizing the distribution network line losses. Among them, the user-side energy storage can be called every day, and the call of the line interconnection switch is limited by the number of times. To this end, it can be further optimized, and the power flow in the distribution network area G becomes:

[0053] in, Expressed as the current change in branch n at time t.

[0054] Accordingly, the total line loss in the distribution network area G is Can be adjusted accordingly:

[0055] in, It is represented by the resistance after the adjustment of the nth branch structure.

[0056] Accordingly, in step S4 of the distribution network reconstruction method of the present invention, it is necessary to take the minimum equivalent cost of dynamic line loss management as the goal, consider multiple constraints, and construct a distribution network reconstruction optimization model. That is to say, in the present invention, it aims to optimize the distribution network line loss with the minimum equivalent cost; wherein, the minimum equivalent cost refers to the cost level after the line loss is reduced by taking measures such as switching the grid connection switch and calling the user-side energy storage call fee. For example, the equivalent cost f is:

[0057] Among them, p represents the electricity price level of the large power grid; C1 represents the cost of switching the line interconnection switch; C2 represents the cost of calling the user-side energy storage.

[0058]

[0059] Where m represents the number of switching tie switches; p1 represents the cost of switching a tie switch; P soc Indicates the amount of energy storage electricity called up on the user side; p2 indicates the energy storage subsidy cost per kilowatt-hour.

[0060] It should be pointed out that in this embodiment, if the goal is to minimize the equivalent cost of dynamic line loss management, it is necessary to consider the power flow constraint, node voltage constraint, branch capacity constraint, reconstruction number constraint, and energy storage call constraint at the same time: 1. System power flow constraints Ensure real-time power balance of the power grid, that is, the user's power demand is equal to the sum of the power supply from the power grid and the energy storage:

[0061] Among them, P grid,nt , P soc,nt They represent the power supply power of the power grid and the energy storage power supply power of branch n at time t respectively.

[0062] 2. Node voltage constraints The node voltage should not exceed its upper and lower limits:

[0063] Where U nmin , U nmax , U nt They represent the lower limit, upper limit and node voltage at time t of branch n respectively.

[0064] 3. Branch capacity constraints The power flowing through each branch should not exceed the maximum value allowed by its line capacity:

[0065] Where P nmax They respectively represent the maximum line capacity allowed for branch n.

[0066] 4. Reconstruction frequency constraints

[0067] In the formula, , Indicates the maximum number of times the tie switch is started and closed.

[0068] 5. Energy storage call constraints The energy storage charge state does not exceed the rated capacity, and charging and discharging are not performed at the same time.

[0069]

[0070] Among them, SOC t represents the energy storage capacity at time t; E sn Indicates the rated capacity of energy storage; P std With P stc They represent the charging and discharging capacity of the energy storage at time t respectively.

[0071] Accordingly, in step S5 of the distribution network reconstruction method of the present invention, combined with the dynamic prediction results of the power flow distribution, the distribution network reconstruction and energy storage call rules are strictly followed. Under all constraints, the MISOCP optimization algorithm can be specifically applied to solve the distribution network reconstruction optimization model to obtain a variety of distribution network reconstruction optimization strategies. With the minimum equivalent cost of the distribution network reconstruction optimization as the ultimate goal, the optimal distribution network reconstruction strategy is determined by comparison, and the optimal distribution network reconstruction strategy can include: energy storage system call strategy, branch tie switch switching strategy, large power grid power supply strategy, etc.

[0072] At the same time, in this embodiment, after the optimal distribution network reconstruction strategy is output and the corresponding switch operation instructions are issued, the operation status of the distribution network can be monitored in real time, and the reconstruction plan can be adjusted in real time according to the feedback.

[0073] For ease of understanding and application, the present invention also discloses Figure 3 and Figure 4 A distribution network reconstruction system considering dynamic line loss management is shown. The distribution network reconstruction system can be specifically applied to the above-mentioned distribution network reconstruction method of the present invention, which may include: a data input and preprocessing module, a branch flow analysis and prediction module, a dynamic line loss calculation and evaluation module, a distribution network reconstruction optimization module and a decision implementation and monitoring feedback module.

[0074] It should be noted that, in this embodiment, the above-mentioned data input and preprocessing module is used to obtain distribution network data and perform preprocessing; The above-mentioned branch power flow analysis and prediction module is used to construct a line power flow distribution prediction model based on deep learning, which is trained based on the preprocessed distribution network data, and then the trained line power flow distribution prediction model is used to carry out power flow distribution calculation to output the line power flow distribution prediction result; The above-mentioned dynamic line loss calculation and evaluation module is used to construct a dynamic line loss evaluation model for the distribution network, so as to obtain the dynamic line loss of each flow direction of the distribution network flow distribution based on the power flow distribution prediction result; The distribution network reconstruction optimization module is used to construct a distribution network reconstruction optimization model with the goal of minimizing the equivalent cost of dynamic line loss management and considering multiple constraints; The above-mentioned decision implementation and monitoring feedback module is used to solve the distribution network reconstruction optimization model to obtain all distribution network reconstruction strategies that meet all constraints, select and output the optimal distribution network reconstruction strategy; at the same time, the decision implementation and monitoring feedback module can also monitor the operating status of the distribution network in real time, and adjust the distribution network reconstruction strategy in real time according to the feedback.

[0075] From the above, it can be seen that the distribution network reconstruction method and system designed by the present invention can effectively solve the technical problems existing in the existing distribution network reconstruction method. It can not only meet the dynamic line loss management needs of the distribution network under the condition of random load changes in each branch at each moment, but also ensure the reliable operation of the distribution system and the minimum equivalent cost of network reconstruction. It can significantly improve the operating efficiency of the distribution network, reduce power loss and operating costs, and effectively promote energy conservation and emission reduction and technological innovation in the power industry. It has good promotion prospects and application value.

[0076] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A distribution network reconstruction method considering dynamic line loss management, characterized in that: Includes steps: S1: Obtain distribution network data and perform preprocessing; S2: Construct a line flow distribution prediction model based on deep learning, train it based on the preprocessed distribution network data, and then use the trained line flow distribution prediction model to carry out flow distribution calculation to output the line flow distribution prediction result; S3: constructing a distribution network dynamic line loss assessment model to obtain the dynamic line loss of each flow direction of the distribution network flow distribution based on the flow distribution prediction result; S4: Taking the minimization of the equivalent cost of dynamic line loss management as the goal, considering multiple constraints, a distribution network reconstruction optimization model is constructed; S5: Solve the distribution network reconstruction optimization model to obtain all distribution network reconstruction strategies that meet all constraints, select and output the optimal distribution network reconstruction strategy.

2. A distribution network reconstruction method considering dynamic line loss management according to claim 1, characterized in that: In step S1, the distribution network data includes: Distribution network structure, branch parameters, node data, user historical load data, and temperature, humidity, working day and holiday data consistent with the time scale of user historical load data.

3. A distribution network reconstruction method considering dynamic line loss management according to claim 2, characterized in that: In step S1, the preprocessing includes: A local outlier factor detection algorithm is used to identify abnormal load data and missing value data in the user's historical load data, and a sliding average method is used for processing.

4. A distribution network reconstruction method considering dynamic line loss management according to claim 1, characterized in that: In step S2, the steps are specifically included: S21: acquiring load sequence data based on the preprocessed distribution network data, normalizing the load sequence data, and dividing the normalized load sequence data into two parts according to the time series, one part as a training set and the other part as a test set; S22: Construct a line flow distribution prediction model based on deep learning; S23: inputting the training set into the line power flow distribution prediction model to perform model training; S24: Inputting the test set into the trained line power distribution prediction model to output the line power distribution prediction result.

5. A distribution network reconstruction method considering dynamic line loss management according to claim 4, characterized in that: In step S22, a branch prediction network is established using a GRU network, and all branch prediction networks form the line power flow distribution prediction model, while sharing the total load characteristics and linking all branch prediction networks together.

6. A distribution network reconstruction method considering dynamic line loss management according to claim 4, characterized in that: Step S24 is specifically as follows: The test set is input into the trained line flow distribution prediction model to obtain the load prediction result at time t by denormalization. , and obtain the load forecast result of branch n at time t , to obtain the power flow of branch n at time t ; in, is the power flow of branch n at time t; is the voltage of branch n at time t.

7. A distribution network reconstruction method considering dynamic line loss management according to claim 1, characterized in that: In step S4, the constraints include: power flow constraints, node voltage constraints, branch capacity constraints, reconstruction times constraints, and energy storage call constraints.

8. A distribution network reconstruction method considering dynamic line loss management according to claim 1, characterized in that: In step S5, specifically: The MISOCP optimization algorithm is applied to solve the distribution network reconstruction optimization model to obtain all distribution network reconstruction strategies that meet all constraints, and the optimal distribution network reconstruction strategy is selected and output.

9. A distribution network reconstruction system considering dynamic line loss management, characterized in that: The method is applied to the distribution network reconstruction method according to any one of claims 1 to 8, comprising: Data input and preprocessing module, which is used to obtain distribution network data and perform preprocessing; The branch power flow analysis and prediction module is used to construct a line power flow distribution prediction model based on deep learning, which is trained based on the preprocessed distribution network data, and then uses the trained line power flow distribution prediction model to carry out power flow distribution calculation to output the line power flow distribution prediction result; A dynamic line loss calculation and evaluation module, which is used to construct a dynamic line loss evaluation model for the distribution network, so as to obtain the dynamic line loss of each flow of the distribution network flow distribution based on the flow distribution prediction result; The distribution network reconstruction optimization module is used to construct a distribution network reconstruction optimization model with the goal of minimizing the equivalent cost of dynamic line loss management and considering multiple constraints; The decision implementation and monitoring feedback module is used to solve the distribution network reconstruction optimization model to obtain all distribution network reconstruction strategies that meet all constraints, select and output the optimal distribution network reconstruction strategy.

10. A distribution network reconstruction system considering dynamic line loss management according to claim 9, characterized in that: The decision implementation and monitoring feedback module can also monitor the operating status of the distribution network in real time, and adjust the distribution network reconstruction strategy in real time according to the feedback.

Images