Power distribution network fault recovery method, system and equipment based on graph calculation and medium

Abstract

The invention discloses a power distribution network fault recovery method, system and device based on graph calculation and a medium. The method comprises the steps of obtaining operation monitoring data of a fault power distribution network in real time; processing a data structure chart of the power distribution network and the operation monitoring data according to a preset first neural network to obtain a circuit breaker action matrix; processing the data structure chart and the operation monitoring data according to a preset second neural network to obtain a distributed power supply action matrix; according to the circuit breaker action matrix and the distributed power supply action matrix, calculating to obtain a recovery action matrix of the fault power distribution network; and adjusting a circuit breaker in the fault power distribution network according to the circuit breaker action matrix, and adjusting a distributed power supply of the fault power distribution network according to the recovery action matrix to complete fault recovery of the power distribution network. According to the invention, through cooperative processing of the circuit breaker action matrix and the distributed power supply action matrix, the efficiency and accuracy of simultaneous processing of discrete actions of the circuit breaker and continuous adjustment of the distributed power supply are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of distribution network fault recovery, and in particular relates to a distribution network fault recovery method, system, equipment and medium based on graph computing. Background Art

[0002] In the existing active distribution network fault recovery technology, when facing the recovery problem after a distribution network fault occurs, the opening and closing states of the circuit breakers and the output power of the distributed power sources are usually adjusted to achieve rapid recovery from the fault. However, as a key control device in the distribution network, the opening and closing states of the circuit breakers are discrete, while the output power regulation of the distributed power sources is continuous. Traditional methods often find it difficult to effectively integrate these two different types of variables, resulting in limited effectiveness of the fault recovery strategy.

[0003] Existing fault recovery methods based on traditional optimization algorithms (such as genetic algorithms, particle swarm optimization, etc.) also have many shortcomings when dealing with complex optimization problems involving discrete and continuous variables. They cannot fully utilize the electrical characteristics and topological structure information of the distribution network, resulting in low computational efficiency and poor recovery effect. In addition, they are not robust enough when facing changes in the distribution network structure and noisy data. Summary of the invention

[0004] In response to the deficiencies in the prior art, the present application provides a distribution network fault recovery method, system, device and medium based on graph computing to improve the efficiency and accuracy of simultaneously processing discrete actions of circuit breakers and continuous regulation of distributed power sources during distribution network fault recovery.

[0005] To achieve the above objectives, in a first aspect, the present invention provides a distribution network fault recovery method based on graph computing, comprising:

[0006] Obtain operation monitoring data of faulty distribution networks in real time;

[0007] Processing the data structure diagram of the distribution network and the operation monitoring data according to a preset first neural network to obtain a circuit breaker action matrix; wherein the first neural network is a graph neural network pre-trained according to the historical circuit breaker action matrix; and the data structure diagram is constructed according to the topological structure and electrical characteristic information of the distribution network;

[0008] Processing the data structure diagram and the operation monitoring data according to a preset second neural network to obtain a distributed power supply action matrix; wherein the second neural network is a graph neural network pre-trained according to a historical distributed power supply action matrix;

[0009] Calculate the restoration action matrix of the fault distribution network according to the circuit breaker action matrix and the distributed power source action matrix;

[0010] According to the circuit breaker action matrix, the circuit breakers in the faulty distribution network are adjusted, and according to the restoration action matrix, the distributed power sources of the faulty distribution network are adjusted to complete the distribution network fault restoration.

[0011] Compared with the prior art, the embodiments of the present application have the following beneficial effects: real-time acquisition of operation monitoring data of the faulty distribution network, ensuring the accuracy and availability of the data, providing high-quality input data for subsequent fault analysis and recovery decisions, and converting the topological structure and electrical characteristics of the distribution network into the form of a data structure graph. It can provide the necessary network topology and electrical characteristics information for subsequent fault detection, location and recovery strategies, and also facilitate computers to use graph calculations to analyze fault impacts and optimize recovery strategies, thereby improving the efficiency and accuracy of fault recovery. By predicting the circuit breaker action matrix through a trained graph neural network, it can better adapt to the complexity and dynamic changes of the distribution network and quickly predict the optimal circuit breaker operation sequence. , helping to restore the normal operation of the power grid, using the trained graph neural network to predict the best response measures of distributed power sources, ensuring that distributed power sources can play an active role in the fault recovery process, and also providing a basis for the prediction of subsequent coordinated operations of circuit breakers and distributed power sources. The recovery action matrix is ​​calculated according to the circuit breaker action matrix and the distributed power source action matrix. The actions of circuit breakers and distributed power sources are comprehensively considered to generate an overall recovery plan, which ensures the consistency and coordination of the recovery operation and avoids the suboptimal solution that may be caused by a single strategy. The circuit breakers in the faulty distribution network are adjusted according to the circuit breaker action matrix, and the distributed power sources in the faulty distribution network are adjusted according to the recovery action matrix, achieving fast and effective fault recovery.

[0012] In some embodiments of the first aspect of the present application, the real-time acquisition of operation monitoring data of the faulty distribution network includes:

[0013] Collecting original operation monitoring data of the faulty distribution network; the original operation monitoring information includes circuit breaker status, distributed power supply output, node voltage, line current and load demand;

[0014] Preprocessing the original operation monitoring data to obtain the operation monitoring data; wherein the preprocessing includes one or more combinations of the following: data cleaning processing, formatting processing and standardization processing.

[0015] Compared with the prior art, the above embodiment has the following beneficial effects: collecting real-time operating data of the faulty distribution network, and performing pre-processing operations such as cleaning, formatting and standardization on the data to ensure the accuracy and availability of the data, and provide high-quality input data for subsequent fault analysis and recovery decisions.

[0016] In some embodiments of the first aspect of the present application, the data structure diagram is constructed based on the topological structure and electrical characteristic information of the distribution network, including:

[0017] According to the topological structure of the distribution network, obtain the equipment information, equipment location information, equipment electrical characteristic information, transmission line information and connection relationship between each device in the distribution network;

[0018] Each device in the distribution network is defined as a node of the data structure diagram, and the attributes of each node include the device information, device electrical characteristic information and device location information;

[0019] The connection relationship between the devices is defined as an edge of a data structure graph, and the attributes of the edge include the transmission line information.

[0020] Compared with the prior art, the above embodiment has the following beneficial effects: by converting the topological structure and electrical characteristics of the distribution network into the form of a data structure graph, it can provide the necessary network topology and electrical characteristics information for subsequent fault detection, location and recovery strategies, and it also facilitates computers to use graph calculations to analyze fault impacts and optimize recovery strategies, thereby improving the efficiency and accuracy of fault recovery.

[0021] In some embodiments of the first aspect of the present application, the first neural network is a graph neural network pre-trained according to a historical circuit breaker action matrix, comprising:

[0022] Obtain the historical data structure diagram and historical operation monitoring data of the historical fault distribution network;

[0023] In the distribution network with historical faults, fault recovery is performed only by adjusting the circuit breakers, and the historical circuit breaker action matrix is ​​obtained;

[0024] The historical data structure diagram and the historical operation monitoring data are used as input, and the historical circuit breaker action matrix is ​​used as output, and the graph neural network is trained to obtain the first neural network.

[0025] Compared with the prior art, the above embodiment has the following beneficial effects: the graph neural network is trained using historical data structure diagrams and historical operation monitoring data to help the graph neural network understand the patterns and relationships of the structure diagram; after training using the historical circuit breaker action matrix, the circuit breaker action under the current fault condition can be predicted, and the optimal operation sequence of the circuit breaker currently used for fault recovery can be output, providing data-driven decision support for subsequent fault recovery.

[0026] In some embodiments of the first aspect of the present application, the second neural network is a graph neural network pre-trained according to a historical distributed power source action matrix, comprising:

[0027] In the distribution network with historical faults, fault recovery is performed only by adjusting the distributed generation, and the historical distributed generation action matrix is ​​obtained;

[0028] The historical data structure diagram and the historical operation monitoring data are used as input, and the historical distributed power supply action matrix is ​​used as output, and the graph neural network is trained to obtain the second neural network.

[0029] Compared with the prior art, the above-mentioned embodiment has the following beneficial effects: the graph neural network is trained using historical data structure diagrams and historical operation monitoring data to help the graph neural network understand the patterns and relationships of the structure diagram; after training using the historical distributed power supply action matrix, the distributed power supply action under the current fault condition can be predicted, and the distributed power supply action matrix currently used for fault recovery can be output, ensuring that the distributed power supply can play an active role in the fault recovery process, while also providing a basis for predicting the subsequent coordinated operation of circuit breakers and distributed power supplies.

[0030] In some embodiments of the first aspect of the present application, the calculating, according to the circuit breaker action matrix and the distributed power source action matrix, a restoration action matrix of the fault distribution network includes:

[0031] Setting a circuit breaker weight matrix for the circuit breaker action matrix, and setting a distributed power source weight matrix for the distributed power source action matrix;

[0032] According to the connection relationship between the devices, the connection relationship between the circuit breaker and the distributed power source is obtained, and according to the connection relationship between the circuit breaker and the distributed power source, a connection relationship matrix is ​​set;

[0033] Calculating an original restoration action matrix according to the circuit breaker weight matrix, the distributed power source weight matrix and the connection relationship matrix;

[0034] The original restored action matrix is ​​normalized to obtain a restored action matrix.

[0035] Compared with the prior art, the above embodiment has the following beneficial effects: by comprehensively considering the actions of circuit breakers and distributed power sources, an overall recovery plan is generated, which can achieve more refined and differentiated fault recovery control, while also ensuring the consistency and coordination of recovery operations, avoiding the suboptimal solution that may be caused by a single strategy.

[0036] In some embodiments of the first aspect of the present application, adjusting the circuit breaker in the faulty distribution network according to the circuit breaker action matrix, adjusting the distributed power supply of the faulty distribution network according to the recovery action matrix, and completing the distribution network fault recovery includes:

[0037] The adjusted power of the distributed power source is calculated according to the elements of the restoration action matrix, and the distributed power source is adjusted according to the adjusted power.

[0038] Compared with the prior art, the above embodiment has the following beneficial effects: according to the calculation results of the circuit breaker action matrix, the circuit breakers in the faulty distribution network are adjusted, and according to the calculation results of the recovery action matrix, the distributed power sources in the faulty distribution network are adjusted. A data-driven approach is used to accurately indicate whether each circuit breaker should be closed or disconnected and how the distributed power sources should adjust their output power, so as to realize the recovery of the faulty distribution network. This not only improves the automation level and efficiency of fault recovery, but also reduces the errors that may be caused by human intervention, thereby ensuring the accuracy of fault recovery.

[0039] In a second aspect, an embodiment of the present application further provides a distribution network fault recovery system based on graph computing, comprising: a monitoring module, a first neural module, a second neural module, a recovery matrix calculation module and a fault recovery module;

[0040] Wherein, the monitoring module is used to obtain the operation monitoring data of the fault distribution network in real time;

[0041] The first neural module is used to process the data structure diagram of the distribution network and the operation monitoring data according to a preset first neural network to obtain a circuit breaker action matrix; wherein the first neural network is a graph neural network pre-trained according to a historical circuit breaker action matrix; the data structure diagram is constructed based on the topological structure and electrical characteristic information of the distribution network;

[0042] The second neural module is used to process the data structure diagram and the operation monitoring data according to a preset second neural network to obtain a distributed power supply action matrix; wherein the second neural network is a graph neural network pre-trained according to a historical distributed power supply action matrix;

[0043] The restoration matrix calculation module is used to calculate the restoration action matrix of the fault distribution network according to the circuit breaker action matrix and the distributed power source action matrix;

[0044] The fault recovery module is used to adjust the circuit breakers in the faulty distribution network according to the circuit breaker action matrix, and to adjust the distributed power sources of the faulty distribution network according to the recovery action matrix to complete the distribution network fault recovery.

[0045] In some embodiments of the second aspect of the present application, the monitoring module includes: a collection unit and a preprocessing unit;

[0046] The acquisition unit is used to collect original operation monitoring data of the distribution network; the original operation monitoring information includes circuit breaker status, distributed power supply output, node voltage, line current and load demand;

[0047] The preprocessing unit is used to preprocess the original operation monitoring data to obtain the operation monitoring data; wherein the preprocessing includes one or more combinations of the following: data cleaning processing, formatting processing and standardization processing.

[0048] In some embodiments of the second aspect of the present application, the data structure diagram is constructed based on the topological structure and electrical characteristic information of the distribution network, including:

[0049] According to the topological structure of the distribution network, obtain the equipment information, equipment location information, equipment electrical characteristic information, transmission line information and connection relationship between each device in the distribution network;

[0050] Each device in the distribution network is defined as a node of the data structure diagram, and the attributes of each node include the device information, device electrical characteristic information and device location information;

[0051] The connection relationship between the devices is defined as an edge of a data structure graph, and the attributes of the edge include the transmission line information.

[0052] In some embodiments of the second aspect of the present application, the first neural network is a graph neural network pre-trained according to a historical circuit breaker action matrix, comprising:

[0053] Obtain the historical data structure diagram and historical operation monitoring data of the historical fault distribution network;

[0054] In the distribution network with historical faults, fault recovery is performed only by adjusting the circuit breakers, and the historical circuit breaker action matrix is ​​obtained;

[0055] The historical data structure diagram and the historical operation monitoring data are used as input, and the historical circuit breaker action matrix is ​​used as output, and the graph neural network is trained to obtain the first neural network.

[0056] In some embodiments of the second aspect of the present application, the second neural network is a graph neural network pre-trained according to a historical distributed power source action matrix, including:

[0057] In the distribution network with historical faults, fault recovery is performed only by adjusting the distributed generation, and the historical distributed generation action matrix is ​​obtained;

[0058] The historical data structure diagram and the historical operation monitoring data are used as input, and the historical distributed power supply action matrix is ​​used as output, and the graph neural network is trained to obtain the second neural network.

[0059] In some embodiments of the second aspect of the present application, the restoration matrix calculation module includes: a weight matrix setting unit, a relationship matrix setting unit, a calculation unit and a normalization unit;

[0060] The weight matrix setting unit is used to set a circuit breaker weight matrix for the circuit breaker action matrix, and to set a distributed power source weight matrix for the distributed power source action matrix;

[0061] The relationship matrix setting unit is used to obtain the connection relationship between the circuit breaker and the distributed power source according to the connection relationship between the devices, and set the connection relationship matrix according to the connection relationship between the circuit breaker and the distributed power source;

[0062] The calculation unit is used to calculate the original restoration action matrix according to the circuit breaker weight matrix, the distributed power source weight matrix and the connection relationship matrix;

[0063] The normalization unit is used to perform normalization processing on the original restored action matrix to obtain a restored action matrix.

[0064] In some embodiments of the second aspect of the present application, the fault recovery module includes: a circuit breaker adjustment unit and a power supply adjustment unit;

[0065] The circuit breaker adjustment unit is used to adjust the circuit breaker in the fault distribution network according to the circuit breaker action matrix;

[0066] The power supply adjustment unit is used to calculate the adjustment power of the distributed power supply according to the elements of the restoration action matrix, and adjust the distributed power supply according to the adjustment power.

[0067] Compared with the prior art, the embodiments of the present application have the following beneficial effects: real-time acquisition of operation monitoring data of the faulty distribution network, ensuring the accuracy and availability of the data, providing high-quality input data for subsequent fault analysis and recovery decisions, and converting the topological structure and electrical characteristics of the distribution network into the form of a data structure graph. It can provide the necessary network topology and electrical characteristics information for subsequent fault detection, location and recovery strategies, and also facilitate computers to use graph calculations to analyze fault impacts and optimize recovery strategies, thereby improving the efficiency and accuracy of fault recovery. By predicting the circuit breaker action matrix through a trained graph neural network, it can better adapt to the complexity and dynamic changes of the distribution network and quickly predict the optimal circuit breaker operation sequence. , helping to restore the normal operation of the power grid, using the trained graph neural network to predict the best response measures of distributed power sources, ensuring that distributed power sources can play an active role in the fault recovery process, and also providing a basis for the prediction of subsequent coordinated operations of circuit breakers and distributed power sources. The recovery action matrix is ​​calculated according to the circuit breaker action matrix and the distributed power source action matrix. The actions of circuit breakers and distributed power sources are comprehensively considered to generate an overall recovery plan, which ensures the consistency and coordination of the recovery operation and avoids the suboptimal solution that may be caused by a single strategy. The circuit breakers in the faulty distribution network are adjusted according to the circuit breaker action matrix, and the distributed power sources in the faulty distribution network are adjusted according to the recovery action matrix, achieving fast and effective fault recovery.

[0068] In a third aspect, an embodiment of the present application also provides an electronic device for distribution network fault recovery based on graph computing, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when loaded into the processor, implements the steps of a distribution network fault recovery method based on graph computing.

[0069] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the distribution network fault recovery method based on graph computing are implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 : A flow chart of a distribution network fault recovery method based on graph computing provided in some embodiments of the present invention.

[0071] Figure 2 : A structural schematic diagram of a distribution network fault recovery system based on graph computing provided in some embodiments of the present invention.

[0072] Figure 3 : A structural diagram of an electronic device for distribution network fault recovery based on graph computing provided in some embodiments of the present invention. DETAILED DESCRIPTION

[0073] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0074] Embodiment 1:

[0075] Please refer to Figure 1 , a distribution network fault recovery method based on graph computing provided by an embodiment of the present invention, comprising steps S1 to S5:

[0076] Step S1: Acquire the operation monitoring data of the faulty distribution network in real time.

[0077] In some embodiments of the present application, step S1 can be implemented by the following preferred implementations:

[0078] S11: Collecting original operation monitoring data of the faulty distribution network; the original operation monitoring information includes circuit breaker status, distributed power supply output, node voltage, line current and load demand;

[0079] S12: Preprocessing the original operation monitoring data to obtain operation monitoring data; wherein the preprocessing includes one or more combinations of the following: data cleaning processing, formatting processing and standardization processing.

[0080] In this embodiment, by collecting real-time operating data of the faulty distribution network and performing pre-processing operations such as cleaning, formatting and standardization on the data, the accuracy and availability of the data can be ensured, providing high-quality input data for subsequent fault analysis and recovery decisions.

[0081] Step S2: Processing the data structure diagram of the distribution network and the operation monitoring data according to a preset first neural network to obtain a circuit breaker action matrix.

[0082] Preferably, the data structure diagram can be obtained by the following preferred implementation:

[0083] According to the topological structure of the distribution network, obtain the equipment information, equipment location information, equipment electrical characteristic information, transmission line information and connection relationship between each device in the distribution network;

[0084] Each device in the distribution network is defined as a node of the data structure diagram, and the attributes of each node include the device information, device electrical characteristic information and device location information;

[0085] The connection relationship between the devices is defined as an edge of a data structure graph, and the attributes of the edge include the transmission line information.

[0086] In this preferred embodiment, by converting the topological structure and electrical characteristics of the distribution network into the form of a data structure graph, the necessary network topology and electrical characteristics information can be provided for subsequent fault detection, location and recovery strategies. It also facilitates computers to use graph calculations to analyze fault impacts and optimize recovery strategies, thereby improving the efficiency and accuracy of fault recovery.

[0087] Furthermore, the first neural network can be obtained by pre-training through the following preferred implementations:

[0088] Obtain a historical data structure diagram and historical operation monitoring data of a historical fault distribution network; in the historical fault distribution network, only adjust the circuit breaker to recover from the fault, and obtain a historical circuit breaker action matrix; use the historical data structure diagram and historical operation monitoring data as input, and the historical circuit breaker action matrix as output, to train a graph neural network, and obtain the first neural network.

[0089] In this preferred embodiment, the graph neural network is trained using historical data structure diagrams and historical operation monitoring data to help the graph neural network understand the patterns and relationships of the structure diagram. After training using the historical circuit breaker action matrix, the circuit breaker action under the current fault condition can be predicted, and the optimal circuit breaker operation sequence currently used for fault recovery can be output, providing data-driven decision support for subsequent fault recovery.

[0090] Furthermore, the fault recovery is performed only by adjusting the circuit breaker to obtain the historical circuit breaker action matrix, which can be achieved by the following preferred manner:

[0091] A first constraint condition is set to keep the distributed power source in the faulty distribution network from performing any operation. Under the restriction of the first constraint condition, the fault recovery is performed only by adjusting the circuit breaker to obtain a historical circuit breaker action matrix. The first constraint condition may be as follows:

[0092]

[0093]

[0094] Among them, x i is the current state of circuit breaker i, y i is the original state of circuit breaker i, N is the total number of circuit breakers, K max is the circuit breaker switching operation times threshold; x j is the current state of circuit breaker j. Islanding refers to a part of the power grid that is disconnected from the main power grid due to the opening and closing operations of some devices (such as circuit breakers).

[0095] In this preferred embodiment, when setting the first constraint condition and adjusting the circuit breaker action, the number of switching times and the island setting of the circuit breaker are fully considered. On the one hand, as a key control device in the distribution network, the circuit breaker's opening and closing operation causes certain losses to the device itself. Frequent opening and closing will shorten the service life of the circuit breaker and may even cause equipment failure. Therefore, it is necessary to limit the number of adjustments to avoid unnecessary losses. On the other hand, when adjusting the circuit breaker, it is necessary to consider the impact of the adjustment on the island. It is necessary to reduce the generation of islands and protect certain loads through islands to ensure the reliability of power supply. Therefore, by adjusting the circuit breaker under the preset first constraint condition, a more accurate and effective historical circuit breaker action matrix can be obtained.

[0096] In step S2 of the present application, the circuit breaker action matrix is ​​predicted by the trained graph neural network, which can better adapt to the complexity and dynamic changes of the distribution network, quickly predict the optimal circuit breaker operation sequence, and help restore the normal operation of the power grid.

[0097] Step S3: Processing the data structure diagram and the operation monitoring data according to a preset second neural network to obtain a distributed power supply action matrix.

[0098] Preferably, the second neural network can be pre-trained by the following preferred implementations:

[0099] In a distribution network with historical faults, fault recovery is performed only by adjusting distributed power sources to obtain a historical distributed power source action matrix; the historical data structure diagram and historical operation monitoring data are used as input, and the historical distributed power source action matrix is ​​used as output to train the graph neural network to obtain the second neural network.

[0100] In this preferred embodiment, the graph neural network is trained using historical data structure diagrams and historical operation monitoring data to help the graph neural network understand the patterns and relationships of the structure diagram. After training using the historical distributed power supply action matrix, the distributed power supply action under the current fault condition can be predicted, and the distributed power supply action matrix currently used for fault recovery can be output, providing data-driven decision support for subsequent fault recovery.

[0101] Furthermore, the above-mentioned fault recovery is performed only by adjusting the distributed power supply to obtain the historical distributed power supply action matrix, which can be achieved by the following preferred implementation manner:

[0102] A second constraint condition is set to keep the circuit breaker in the faulty distribution network from performing any operation. Under the restriction of the second constraint condition, the fault recovery is performed only by adjusting the distributed power source to obtain the historical distributed power source action matrix. The second constraint condition can be as follows:

[0103] P DGi,min ≤P DGi ≤P DGi,max ;

[0104]

[0105] V i,min ≤V i ≤V i,max ;

[0106] Among them, P DGi,min is the minimum output power of distributed power source i, P DGi is the output power of distributed power source i, P DGi,max is the maximum output power of distributed power source i, M is the number of distributed power sources, P grid The power provided to the main grid, L j is the load power, P loss is the network loss, V i,min is the minimum output voltage of distributed power source i, V i is the output voltage of distributed power source i, V i,max is the maximum output voltage of distributed power source i.

[0107] In this preferred embodiment, when setting the second constraint condition, multiple factors such as the output power limit of the distributed power source, the output voltage range, and the overall power balance of the distribution network are fully considered. These factors ensure that the operation of the distributed power source in the fault recovery process not only complies with its own physical limitations, but also ensures the stable operation of the power grid. The distributed power source is adjusted under the restriction of the second constraint condition, and the output power and voltage of the distributed power source are precisely controlled. The power flow of the power grid can be adjusted more finely, thereby meeting the load demand while avoiding excessive impact on the power grid. It can also effectively reduce energy losses in the fault recovery process and improve overall energy utilization efficiency.

[0108] In step S3 of this embodiment, the distributed power supply action matrix is ​​predicted by the trained graph neural network, which can ensure that the distributed power supply plays an active role in the fault recovery process, and also provide a basis for the prediction of subsequent coordinated operation of circuit breakers and distributed power supplies.

[0109] Step S4: Calculate the restoration action matrix of the fault distribution network according to the circuit breaker action matrix and the distributed power source action matrix.

[0110] In some embodiments of the present application, step S4 can be implemented by the following preferred implementations:

[0111] S41: setting a circuit breaker weight matrix for the circuit breaker action matrix, and setting a distributed power weight matrix for the distributed power action matrix;

[0112] S42: acquiring a connection relationship between a circuit breaker and a distributed power source according to the connection relationship between the devices, and setting a connection relationship matrix according to the connection relationship between the circuit breaker and the distributed power source;

[0113] S43: Calculate an original restoration action matrix according to the circuit breaker weight matrix, the distributed power source weight matrix and the connection relationship matrix;

[0114] S44: normalizing the original restored action matrix to obtain a restored action matrix.

[0115] Furthermore, the original restoration action matrix is ​​calculated based on the circuit breaker weight matrix, the distributed power source weight matrix and the connection relationship matrix. The calculation algorithm may be as follows:

[0116] Q NM =[q ij ];

[0117] q ij =ω Ni ×B i +ω Mj ×DG j ×a ij ;

[0118] rij=q ij ×a ij ;

[0119] Among them, Q NM is the transition matrix, q ij is the element of the transition matrix, ω Ni is the weight of circuit breaker i, from the circuit breaker weight matrix W N =[ω Ni ], ω Mj is the weight of distributed power j, from the distributed power weight matrix W M =[ω Mj ], B i is the state of circuit breaker i, the value is 0 when it is open and 1 when it is closed. From the circuit breaker action matrix B = [B i ], DG j is the power adjustment coefficient of distributed power j, from the distributed power action matrix DG = [DG j], aij is the connection relationship between the ith circuit breaker and the jth distributed generation, the value is 1 when connected, otherwise it is 0, obtained from the connection relationship matrix ANM = [aij], rij is the element in the original recovery action matrix.

[0120] For example, the circuit breaker action matrix B = [1, 0] output by the first neural network indicates that the first circuit breaker is closed and the second circuit breaker is open;

[0121] The distributed power generation action matrix DG = [0.8, 0.6] output by the second neural network indicates that the output power of the first distributed power generation is 80% and the output power of the second distributed power generation is 60%;

[0122] The circuit breaker weight matrix W can be set based on experience or historical data analysis. N =[0.6, 0.4], distributed power weight matrix W M = [0.7, 0.3];

[0123] Obtained connection relationship matrix a ij =1 means that the i-th circuit breaker is connected to the j-th distributed generation; a ij =0 means that the i-th circuit breaker is not connected to the j-th distributed generation;

[0124] Calculate the transition matrix

[0125] Calculate the original recovery action matrix

[0126] Normalize the original recovery action matrix to obtain the recovery action matrix

[0127] In this embodiment, the actions of circuit breakers and distributed power sources are comprehensively considered to generate an overall recovery plan, which can achieve more refined and differentiated fault recovery control, while also ensuring the consistency and coordination of recovery operations, avoiding the suboptimal solution that may be caused by a single strategy.

[0128] Step S5: According to the circuit breaker action matrix, the circuit breakers in the faulty distribution network are adjusted, and according to the recovery action matrix, the distributed power sources of the faulty distribution network are adjusted to complete the distribution network fault recovery.

[0129] In some embodiments of the present application, step S5 can be implemented by the following preferred implementations:

[0130] S51: adjusting the circuit breakers in the fault distribution network according to the circuit breaker action matrix;

[0131] S52: Calculate the adjusted power of the distributed power source according to the elements of the recovery action matrix, and adjust the distributed power source according to the adjusted power.

[0132] Furthermore, the calculation formula for calculating the adjusted power of the distributed power source may be as follows:

[0133] Among them, P DGj is the adjusted power of distributed generation j.

[0134] For example, assuming that the circuit breaker action matrix B = [1, 0] output by the first neural network, the first circuit breaker is closed, the second circuit breaker is opened, and so on. The circuit breakers are adjusted in this way, and the original recovery action matrix is ​​normalized. Assuming that the recovery action matrix is ​​obtained Then the adjustment power of the distributed power supply is calculated as: P DG =[1,0.16], according to the power adjustment, the output power of the first distributed power source is adjusted to 100%, and the output power of the second distributed power source is adjusted to 16%, and the fault recovery adjustment is completed.

[0135] In this embodiment, the circuit breakers in the faulty distribution network are adjusted according to the calculation results of the circuit breaker action matrix, and the distributed power sources in the faulty distribution network are adjusted according to the calculation results of the recovery action matrix. A data-driven approach is used to accurately indicate whether each circuit breaker should be closed or opened and how the distributed power sources should adjust their output power, so as to realize the recovery of the faulty distribution network. This not only improves the automation level and efficiency of fault recovery, but also reduces the errors that may be caused by human intervention, thereby ensuring the accuracy of fault recovery.

[0136] In summary, compared with the prior art, the embodiments of the present application have the following beneficial effects: real-time acquisition of operation monitoring data of the faulty distribution network, ensuring the accuracy and availability of the data, providing high-quality input data for subsequent fault analysis and recovery decisions, and converting the topological structure and electrical characteristics of the distribution network into the form of a data structure graph. It can provide the necessary network topology and electrical characteristics information for subsequent fault detection, location and recovery strategies, and also facilitate computers to use graph calculations to analyze fault impacts and optimize recovery strategies, thereby improving the efficiency and accuracy of fault recovery. By predicting the circuit breaker action matrix through a trained graph neural network, it can better adapt to the complexity and dynamic changes of the distribution network and quickly predict the optimal circuit breaker operation. Sequence, help restore the normal operation of the power grid, use the trained graph neural network to predict the best response measures of distributed power sources, ensure that distributed power sources can play an active role in the fault recovery process, and also provide a basis for the prediction of subsequent coordinated operations of circuit breakers and distributed power sources. The recovery action matrix is ​​calculated according to the circuit breaker action matrix and the distributed power source action matrix. The actions of circuit breakers and distributed power sources are comprehensively considered to generate an overall recovery plan, which ensures the consistency and coordination of the recovery operation and avoids the suboptimal solution that may be caused by a single strategy. The circuit breakers in the faulty distribution network are adjusted according to the circuit breaker action matrix, and the distributed power sources in the faulty distribution network are adjusted according to the recovery action matrix, thereby achieving fast and effective fault recovery.

[0137] Embodiment 2:

[0138] Please refer to Figure 2 , based on the same inventive concept, a distribution network fault recovery system based on graph computing disclosed in an embodiment of the present invention includes: a monitoring module M1, a first neural module M2, a second neural module M3, a recovery matrix calculation module M4 and a fault recovery module M5;

[0139] The monitoring module M1 is used to obtain the operation monitoring data of the faulty distribution network in real time.

[0140] The monitoring module M1 includes: a collection unit and a preprocessing unit; the collection unit is used to collect original operation monitoring data of the distribution network; the original operation monitoring information includes circuit breaker status, distributed power supply output, node voltage, line current and load demand; the preprocessing unit is used to preprocess the original operation monitoring data to obtain operation monitoring data; wherein, the preprocessing includes one or more combinations of the following: data cleaning processing, formatting processing and standardization processing.

[0141] The monitoring module M1 of this embodiment collects real-time operating data of the faulty distribution network and performs pre-processing operations such as cleaning, formatting and standardization on the data to ensure the accuracy and availability of the data, thereby providing high-quality input data for subsequent fault analysis and recovery decisions.

[0142] The first neural module M2 is used to process the data structure diagram of the distribution network and the operation monitoring data according to a preset first neural network to obtain a circuit breaker action matrix; wherein the first neural network is a graph neural network pre-trained according to historical circuit breaker action matrices; and the data structure diagram is constructed based on the topological structure and electrical characteristic information of the distribution network.

[0143] The first neural module M2 of this embodiment predicts the circuit breaker action matrix through the trained graph neural network, which can better adapt to the complexity and dynamic changes of the distribution network, quickly predict the optimal circuit breaker operation sequence, and help restore the normal operation of the power grid.

[0144] Preferably, the data structure diagram is constructed based on the topological structure and electrical characteristic information of the distribution network, including:

[0145] According to the topological structure of the distribution network, obtain the equipment information, equipment location information, equipment electrical characteristic information, transmission line information and connection relationship between each device in the distribution network;

[0146] Each device in the distribution network is defined as a node of the data structure diagram, and the attributes of each node include the device information, device electrical characteristic information and device location information;

[0147] The connection relationship between the devices is defined as an edge of a data structure graph, and the attributes of the edge include the transmission line information.

[0148] In this preferred embodiment, by converting the topological structure and electrical characteristics of the distribution network into the form of a data structure graph, the necessary network topology and electrical characteristics information can be provided for subsequent fault detection, location and recovery strategies. It also facilitates computers to use graph calculations to analyze fault impacts and optimize recovery strategies, thereby improving the efficiency and accuracy of fault recovery.

[0149] Furthermore, the first neural network is a graph neural network pre-trained according to a historical circuit breaker action matrix, comprising:

[0150] Obtain the historical data structure diagram and historical operation monitoring data of the historical fault distribution network;

[0151] In the distribution network with historical faults, fault recovery is performed only by adjusting the circuit breakers, and the historical circuit breaker action matrix is ​​obtained;

[0152] The historical data structure diagram and the historical operation monitoring data are used as input, and the historical circuit breaker action matrix is ​​used as output, and the graph neural network is trained to obtain the first neural network.

[0153] In this embodiment, the graph neural network is trained using historical data structure diagrams and historical operation monitoring data to help the graph neural network understand the patterns and relationships of the structure diagram. After training using the historical circuit breaker action matrix, the circuit breaker action under the current fault condition can be predicted, and the optimal circuit breaker operation sequence currently used for fault recovery can be output, providing data-driven decision support for subsequent fault recovery.

[0154] The second neural module M3 is used to process the data structure diagram and the operation monitoring data according to a preset second neural network to obtain a distributed power supply action matrix; wherein the second neural network is a graph neural network pre-trained based on a historical distributed power supply action matrix.

[0155] The second neural module M3 of this embodiment predicts the distributed power supply action matrix through the trained graph neural network, which can ensure that the distributed power supply plays an active role in the fault recovery process, and also provides a basis for the prediction of subsequent coordinated operations of circuit breakers and distributed power supplies.

[0156] Preferably, the second neural network is a graph neural network pre-trained according to a historical distributed power generation action matrix, comprising:

[0157] In the distribution network with historical faults, fault recovery is performed only by adjusting the distributed generation, and the historical distributed generation action matrix is ​​obtained;

[0158] The historical data structure diagram and the historical operation monitoring data are used as input, and the historical distributed power supply action matrix is ​​used as output, and the graph neural network is trained to obtain the second neural network.

[0159] In this preferred embodiment, the graph neural network is trained using historical data structure diagrams and historical operation monitoring data to help the graph neural network understand the patterns and relationships of the structure diagram. After training using the historical distributed power supply action matrix, the distributed power supply action under the current fault condition can be predicted, and the distributed power supply action matrix currently used for fault recovery can be output to ensure that the distributed power supply can play an active role in the fault recovery process, while also providing a basis for predicting the subsequent coordinated operation of circuit breakers and distributed power supplies.

[0160] The restoration matrix calculation module M4 is used to calculate the restoration action matrix of the fault distribution network according to the circuit breaker action matrix and the distributed power source action matrix.

[0161] The recovery matrix calculation module M4 includes: a weight matrix setting unit, a relationship matrix setting unit, a calculation unit and a normalization unit; the weight matrix setting unit is used to set a circuit breaker weight matrix for the circuit breaker action matrix, and to set a distributed power supply weight matrix for the distributed power supply action matrix; the relationship matrix setting unit is used to obtain the connection relationship between the circuit breaker and the distributed power supply according to the connection relationship between the devices, and to set a connection relationship matrix according to the connection relationship between the circuit breaker and the distributed power supply; the calculation unit is used to calculate the original recovery action matrix according to the circuit breaker weight matrix, the distributed power supply weight matrix and the connection relationship matrix; the normalization unit is used to normalize the original recovery action matrix to obtain a recovery action matrix.

[0162] The recovery matrix calculation module M4 of this embodiment generates an overall recovery plan by comprehensively considering the actions of circuit breakers and distributed power sources, which can achieve more refined and differentiated fault recovery control, while also ensuring the consistency and coordination of recovery operations, avoiding the suboptimal solution that may be caused by a single strategy.

[0163] The fault recovery module M5 is used to adjust the circuit breakers in the faulty distribution network according to the circuit breaker action matrix, and to adjust the distributed power sources of the faulty distribution network according to the recovery action matrix to complete the distribution network fault recovery.

[0164] The fault recovery module M5 includes: a circuit breaker adjustment unit and a power supply adjustment unit; the circuit breaker adjustment unit is used to adjust the circuit breaker in the fault distribution network according to the circuit breaker action matrix; the power supply adjustment unit is used to calculate the adjustment power of the distributed power source according to the elements of the recovery action matrix, and adjust the distributed power source according to the adjustment power.

[0165] The fault recovery module M5 of this embodiment adjusts the circuit breaker in the faulty distribution network according to the calculation result of the circuit breaker action matrix, and adjusts the distributed power supply in the faulty distribution network according to the calculation result of the recovery action matrix. It uses a data-driven approach to accurately indicate whether each circuit breaker should be closed or opened and how the distributed power supply should adjust the output power, thereby realizing the recovery of the faulty distribution network. This not only improves the automation level and efficiency of fault recovery, but also reduces the errors that may be caused by human intervention, thereby ensuring the accuracy of fault recovery.

[0166] In summary, compared with the prior art, the embodiments of the present application have the following beneficial effects: real-time acquisition of operation monitoring data of the faulty distribution network, ensuring the accuracy and availability of the data, providing high-quality input data for subsequent fault analysis and recovery decisions, and converting the topological structure and electrical characteristics of the distribution network into the form of a data structure graph. It can provide the necessary network topology and electrical characteristics information for subsequent fault detection, location and recovery strategies, and also facilitate computers to use graph calculations to analyze fault impacts and optimize recovery strategies, thereby improving the efficiency and accuracy of fault recovery. By predicting the circuit breaker action matrix through a trained graph neural network, it can better adapt to the complexity and dynamic changes of the distribution network and quickly predict the optimal circuit breaker operation. Sequence, help restore the normal operation of the power grid, use the trained graph neural network to predict the best response measures of distributed power sources, ensure that distributed power sources can play an active role in the fault recovery process, and also provide a basis for the prediction of subsequent coordinated operations of circuit breakers and distributed power sources. The recovery action matrix is ​​calculated according to the circuit breaker action matrix and the distributed power source action matrix. The actions of circuit breakers and distributed power sources are comprehensively considered to generate an overall recovery plan, which ensures the consistency and coordination of the recovery operation and avoids the suboptimal solution that may be caused by a single strategy. The circuit breakers in the faulty distribution network are adjusted according to the circuit breaker action matrix, and the distributed power sources in the faulty distribution network are adjusted according to the recovery action matrix, thereby achieving fast and effective fault recovery.

[0167] The division of the modules described above is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules may be combined or integrated into another system.

[0168] Embodiment three:

[0169] Figure 3 The structure diagram of an electronic device for distribution network fault recovery based on graph computing in this application is presented. Figure 3 As shown, the fault detection device may include: a processor N1, a memory N2, a data interface N3 and a communication bus N4.

[0170] Among them: the processor N1, the memory N2, and the data interface N3 communicate with each other through the communication bus N4; the data interface N3 is used for data communication with other additional devices; the processor N1 is used to execute the program N5, which can specifically execute the relevant steps in the above-mentioned distribution network fault recovery method embodiment based on graph calculation.

[0171] Specifically, the program N5 may include program codes, and the program codes include computer executable instructions.

[0172] The processor N1 may be a central processing unit CPU, or an application-specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the distribution network fault recovery device based on graph computing may be processors of the same type, such as one or more CPUs, or may be processors of different types, such as one or more CPUs and one or more ASICs.

[0173] The memory N2 is used to store the program N5. The memory N2 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0174] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system or other device. In addition, the embodiments of the present application are not directed to any particular programming language.

[0175] Embodiment 4:

[0176] An embodiment of the present invention also provides a computer-readable storage medium, which stores at least one executable instruction. When the executable instruction is executed on a distribution network fault recovery device / system based on graph computing, the distribution network fault recovery device / system based on graph computing executes the distribution network fault recovery method based on graph computing in any of the above method embodiments.

[0177] In the specification provided herein, a large number of specific details are described. However, it is understood that the embodiments of the present application can be practiced without these specific details. Similarly, in order to simplify the application and help understand one or more of the various inventive aspects, in the above description of the exemplary embodiments of the present application, the various features of the embodiments of the present application are sometimes grouped together into a single embodiment, figure, or description thereof. Wherein, the claims that follow the specific embodiment are thus clearly incorporated into the specific embodiment, wherein each claim itself is used as a separate embodiment of the application.

[0178] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and further may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive.

Claims

1. A distribution network fault recovery method based on graph computing, characterized in that: include: Obtain operation monitoring data of faulty distribution networks in real time; Processing the data structure diagram of the distribution network and the operation monitoring data according to a preset first neural network to obtain a circuit breaker action matrix; wherein the first neural network is a graph neural network pre-trained according to the historical circuit breaker action matrix; and the data structure diagram is constructed according to the topological structure and electrical characteristic information of the distribution network; Processing the data structure diagram and the operation monitoring data according to a preset second neural network to obtain a distributed power supply action matrix; wherein the second neural network is a graph neural network pre-trained according to a historical distributed power supply action matrix; Calculate the restoration action matrix of the fault distribution network according to the circuit breaker action matrix and the distributed power source action matrix; According to the circuit breaker action matrix, the circuit breakers in the faulty distribution network are adjusted, and according to the restoration action matrix, the distributed power sources of the faulty distribution network are adjusted to complete the distribution network fault restoration.

2. A distribution network fault recovery method based on graph computing as claimed in claim 1, characterized in that: The real-time acquisition of operation monitoring data of the faulty distribution network includes: Collecting original operation monitoring data of the faulty distribution network; the original operation monitoring information includes circuit breaker status, distributed power supply output, node voltage, line current and load demand; Preprocessing the original operation monitoring data to obtain the operation monitoring data; wherein the preprocessing includes one or more combinations of the following: data cleaning processing, formatting processing and standardization processing.

3. A distribution network fault recovery method based on graph computing as claimed in claim 2, characterized in that: The data structure diagram is constructed based on the topological structure and electrical characteristic information of the distribution network, including: According to the topological structure of the distribution network, obtain the equipment information, equipment location information, equipment electrical characteristic information, transmission line information and connection relationship between each device in the distribution network; Each device in the distribution network is defined as a node of the data structure diagram, and the attributes of each node include the device information, device electrical characteristic information and device location information; The connection relationship between the devices is defined as an edge of a data structure graph, and the attributes of the edge include the transmission line information.

4. A distribution network fault recovery method based on graph computing as claimed in claim 3, characterized in that: The first neural network is a graph neural network pre-trained according to the historical circuit breaker action matrix, including: Obtain the historical data structure diagram and historical operation monitoring data of the historical fault distribution network; In the distribution network with historical faults, fault recovery is performed only by adjusting the circuit breakers, and the historical circuit breaker action matrix is ​​obtained; The historical data structure diagram and the historical operation monitoring data are used as input, and the historical circuit breaker action matrix is ​​used as output, and the graph neural network is trained to obtain the first neural network.

5. A distribution network fault recovery method based on graph computing as claimed in claim 4, characterized in that: The second neural network is a graph neural network pre-trained according to the historical distributed power generation action matrix, including: In the distribution network with historical faults, fault recovery is performed only by adjusting the distributed generation, and the historical distributed generation action matrix is ​​obtained; The historical data structure diagram and the historical operation monitoring data are used as input, and the historical distributed power supply action matrix is ​​used as output, and the graph neural network is trained to obtain the second neural network.

6. The distribution network fault recovery method based on graph computing according to claim 5, characterized in that: The calculating, according to the circuit breaker action matrix and the distributed power source action matrix, a restoration action matrix of the fault distribution network comprises: Setting a circuit breaker weight matrix for the circuit breaker action matrix, and setting a distributed power weight matrix for the distributed power action matrix; According to the connection relationship between the devices, the connection relationship between the circuit breaker and the distributed power source is obtained, and according to the connection relationship between the circuit breaker and the distributed power source, a connection relationship matrix is ​​set; Calculating an original restoration action matrix according to the circuit breaker weight matrix, the distributed power source weight matrix and the connection relationship matrix; The original restored action matrix is ​​normalized to obtain a restored action matrix.

7. The distribution network fault recovery method based on graph computing according to claim 6, characterized in that: The method of adjusting the circuit breaker in the faulty distribution network according to the circuit breaker action matrix and adjusting the distributed power supply of the faulty distribution network according to the recovery action matrix to complete the distribution network fault recovery includes: The adjusted power of the distributed power source is calculated according to the elements of the restoration action matrix, and the distributed power source is adjusted according to the adjusted power.

8. A distribution network fault recovery system based on graph computing, characterized in that: include: A monitoring module, a first neural module, a second neural module, a recovery matrix calculation module and a fault recovery module; Wherein, the monitoring module is used to obtain the operation monitoring data of the fault distribution network in real time; The first neural module is used to process the data structure diagram of the distribution network and the operation monitoring data according to a preset first neural network to obtain a circuit breaker action matrix; wherein the first neural network is a graph neural network pre-trained according to a historical circuit breaker action matrix; the data structure diagram is constructed based on the topological structure and electrical characteristic information of the distribution network; The second neural module is used to process the data structure diagram and the operation monitoring data according to a preset second neural network to obtain a distributed power supply action matrix; wherein the second neural network is a graph neural network pre-trained according to a historical distributed power supply action matrix; The restoration matrix calculation module is used to calculate the restoration action matrix of the fault distribution network according to the circuit breaker action matrix and the distributed power source action matrix; The fault recovery module is used to adjust the circuit breakers in the faulty distribution network according to the circuit breaker action matrix, and to adjust the distributed power sources of the faulty distribution network according to the recovery action matrix to complete the distribution network fault recovery.

9. The distribution network fault recovery system based on graph computing according to claim 8, characterized in that: The monitoring module includes: a collection unit and a pre-processing unit; The acquisition unit is used to collect the original operation monitoring data of the fault distribution network; the original operation monitoring information includes circuit breaker status, distributed power supply output, node voltage, line current and load demand; The preprocessing unit is used to preprocess the original operation monitoring data to obtain the operation monitoring data; wherein the preprocessing includes one or more combinations of the following: data cleaning processing, formatting processing and standardization processing.

10. The distribution network fault recovery system based on graph computing according to claim 9, characterized in that: The data structure diagram is constructed based on the topological structure and electrical characteristic information of the distribution network, including: According to the topological structure of the distribution network, obtain the equipment information, equipment location information, equipment electrical characteristic information, transmission line information and connection relationship between each device in the distribution network; Each device in the distribution network is defined as a node of the data structure diagram, and the attributes of each node include the device information, device electrical characteristic information and device location information; The connection relationship between the devices is defined as an edge of a data structure graph, and the attributes of the edge include the transmission line information.

11. The distribution network fault recovery system based on graph computing according to claim 10, characterized in that: The first neural network is a graph neural network pre-trained according to the historical circuit breaker action matrix, including: Obtain the historical data structure diagram and historical operation monitoring data of the historical fault distribution network; In the distribution network with historical faults, fault recovery is performed only by adjusting the circuit breakers, and the historical circuit breaker action matrix is ​​obtained; The historical data structure diagram and the historical operation monitoring data are used as input, and the historical circuit breaker action matrix is ​​used as output, and the graph neural network is trained to obtain the first neural network.

12. The distribution network fault recovery system based on graph computing according to claim 11, characterized in that: The second neural network is a graph neural network pre-trained according to the historical distributed power generation action matrix, including: In the distribution network with historical faults, fault recovery is performed only by adjusting the distributed generation, and the historical distributed generation action matrix is ​​obtained; The historical data structure diagram and the historical operation monitoring data are used as input, and the historical distributed power supply action matrix is ​​used as output, and the graph neural network is trained to obtain the second neural network.

13. The distribution network fault recovery system based on graph computing according to claim 12, characterized in that: The recovery matrix calculation module includes: a weight matrix setting unit, a relationship matrix setting unit, a calculation unit and a normalization unit; The weight matrix setting unit is used to set a circuit breaker weight matrix for the circuit breaker action matrix, and to set a distributed power source weight matrix for the distributed power source action matrix; The relationship matrix setting unit is used to obtain the connection relationship between the circuit breaker and the distributed power source according to the connection relationship between the devices, and set the connection relationship matrix according to the connection relationship between the circuit breaker and the distributed power source; The calculation unit is used to calculate the original restoration action matrix according to the circuit breaker weight matrix, the distributed power source weight matrix and the connection relationship matrix; The normalization unit is used to perform normalization processing on the original restored action matrix to obtain a restored action matrix.

14. The distribution network fault recovery system based on graph computing according to claim 13, characterized in that: The fault recovery module includes: a circuit breaker adjustment unit and a power supply adjustment unit; The circuit breaker adjustment unit is used to adjust the circuit breaker in the fault distribution network according to the circuit breaker action matrix; The power supply adjustment unit is used to calculate the adjustment power of the distributed power supply according to the elements of the restoration action matrix, and adjust the distributed power supply according to the adjustment power.

15. An electronic device for distribution network fault recovery based on graph computing, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is loaded into the processor, the steps of the distribution network fault recovery method based on graph computing according to any one of claims 1 to 7 are implemented.

16. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a distribution network fault recovery method based on graph computing according to any one of claims 1 to 7 are implemented.

Images