Method and device for determining distributed power supply recovery mode of power distribution network

Abstract

The invention relates to a method and a device for determining a distributed power supply recovery mode of a power distribution network. The method comprises the following steps: acquiring a plurality of sub-regions of a pre-divided active power distribution system; for each sub-region, determining a switching action path representing a topological state of each stage of power supply recovery based on a target power supply recovery model constructed based on a Monte Carlo tree and a minimum load value loss; and according to the switching action path and the target power supply recovery model, performing power supply recovery mode solving processing on the sub-region to obtain a target recovery mode. According to the Monte Carlo tree search, different topological states are continuously explored and evaluated, the switching action sequence problem in the power supply recovery process is effectively solved, and the computing power bottleneck faced when a traditional method is used for processing large-scale non-convex problems is avoided. The target power supply recovery model is used for solving the power supply recovery mode, and the obtained target recovery mode has good optimization performance.

Description

Technical Field

[0001] The present application relates to the technical field of active power distribution systems, and in particular to a method and device for determining a distributed power supply restoration mode of a distribution network. Background Art

[0002] Electricity supply is vital to social stability and economic development. However, due to factors such as extreme weather, natural disasters, and cyber attacks, distribution system failures are still difficult to avoid, which can lead to large-scale power outages and reduce the reliability of the power system. Therefore, it is particularly important to quickly develop a feasible power restoration strategy after a failure occurs.

[0003] The current power restoration algorithm only needs to communicate with neighboring areas to achieve power restoration, which reduces the communication pressure. However, when dealing with complex and high-dimensional decision-making problems, it cannot make good path decisions. Summary of the invention

[0004] Based on this, it is necessary to provide a method and device for determining a distributed power supply restoration mode of a distribution network that can solve the above-mentioned technical problems.

[0005] In a first aspect, the present application provides a method for determining a distributed power supply restoration mode of a distribution network, the method comprising:

[0006] Acquire a plurality of pre-divided sub-areas of an active power distribution system;

[0007] For each sub-area, the switch action path is determined based on the pre-built Monte Carlo tree and the target power restoration model; the nodes of the Monte Carlo tree represent the topological state of the active distribution system, the switch action path represents the topological state of each stage of power restoration, and the target power restoration model is constructed based on the goal of minimizing the load value loss;

[0008] According to the switch action path and the target power supply restoration model, the power supply restoration mode of the sub-area is solved and processed to obtain the target restoration mode.

[0009] In one embodiment, the determining of the switch action path based on the pre-built Monte Carlo tree and the target power restoration model includes:

[0010] In each round of iteration, the path exploration process is performed based on the Monte Carlo tree to obtain the leaf nodes to be processed;

[0011] Perform simulation operation processing on the leaf node to be processed, and determine the value of the leaf node to be processed according to the processing result and the target power supply recovery model;

[0012] The Monte Carlo tree is obtained according to the value update of the leaf node to be processed, and the iteration is terminated when the number of iterations meets the preset conditions, and the search path of the Monte Carlo tree is determined as the switch action path.

[0013] In one embodiment, the path exploration process based on the Monte Carlo tree to obtain the leaf node to be processed includes:

[0014] For any node in the Monte Carlo tree, if the node has not been processed by path exploration, the node is expanded to obtain multiple child nodes;

[0015] Calculate the upper confidence bound value of each child node; the upper confidence bound value is used to evaluate the estimated value of each child node;

[0016] The child node with the maximum upper confidence bound value is determined as the leaf node to be processed.

[0017] In one embodiment, the node is expanded to obtain multiple child nodes, including:

[0018] The topological states of the nodes are enumerated to obtain multiple initial switch action plans;

[0019] Determine multiple feasible switching action plans according to each initial switching action plan and predetermined state constraints;

[0020] Each feasible switch action plan is executed on the node respectively, and the topological state after each feasible switch action plan is executed is determined as a child node.

[0021] In one embodiment, the simulated operation processing is performed on the leaf node to be processed, and the value of the leaf node to be processed is determined according to the processing result and the target power supply recovery model, including:

[0022] Get the initial parameters of the leaf node to be processed;

[0023] Input the initial parameters into the target power supply restoration model to solve the power supply problem and obtain the load value restoration amount of the leaf node to be processed;

[0024] The load value recovery amount of the leaf node to be processed is input into the pre-built value calculation model to perform load value calculation processing to obtain the value of the leaf node to be processed.

[0025] In one embodiment, the process of constructing the target power restoration model includes:

[0026] Based on the pre-acquired distribution system parameter information, an initial power supply restoration model is constructed;

[0027] According to the initial power supply restoration model and the solution model, a target power supply restoration model is constructed.

[0028] In one embodiment, the power distribution system parameter information includes at least one of a plurality of nodes in the sub-area, a total active load of each node, a total reactive load, and a load importance weight value; based on the pre-acquired power distribution system parameter information, an initial power supply restoration model is constructed, including:

[0029] According to the distribution system parameter information, determine the objective function and the first constraint condition; the first constraint condition includes safety operation constraint, power flow constraint and distributed power supply operation constraint; the objective function is based on the total active load, load importance weight value and load loss rate; the objective function is used to characterize the priority of different loads in power supply restoration;

[0030] The initial power supply restoration model is composed of the objective function and the first constraint condition.

[0031] In one embodiment, the target power restoration model is constructed according to the initial power restoration model and the solution model, including:

[0032] Based on the node topology state of the Monte Carlo tree, the initial power supply restoration model is assigned variables to obtain the assigned initial power supply restoration model;

[0033] The target power supply restoration model is composed of the assigned initial power supply restoration model and the solution model.

[0034] In one embodiment, the process of constructing the above-mentioned solution model includes:

[0035] Decompose the objective function to obtain multiple sub-functions; the sub-functions are composed of iteration variables and dual variables;

[0036] Constructing convergence conditions based on pre-acquired convergence parameters; the convergence parameters include the current boundary variables and the previous K boundary variables of the iteration variables;

[0037] The solution model is composed of the pre-acquired second constraint condition, each sub-function and convergence condition.

[0038] In a second aspect, the present application also provides a device for determining a distributed power supply restoration mode of a distribution network. The device comprises:

[0039] A region acquisition module, used for acquiring a plurality of pre-divided sub-regions of an active power distribution system;

[0040] The path determination module is used to determine the switch action path for each sub-area based on the pre-built Monte Carlo tree and the target power supply restoration model; the nodes of the Monte Carlo tree represent the topological state of the active distribution system, the switch action path represents the topological state of each stage of power supply restoration, and the target power supply restoration model is constructed based on the goal of minimizing the load value loss;

[0041] The solution module is used to solve the power supply restoration mode of the sub-area according to the switch action path and the target power supply restoration model to obtain the target restoration mode.

[0042] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0043] Acquire a plurality of pre-divided sub-areas of an active power distribution system;

[0044] For each sub-area, the switch action path is determined based on the pre-built Monte Carlo tree and the target power restoration model; the nodes of the Monte Carlo tree represent the topological state of the active distribution system, the switch action path represents the topological state of each stage of power restoration, and the target power restoration model is constructed based on the goal of minimizing the load value loss;

[0045] According to the switch action path and the target power supply restoration model, the power supply restoration mode of the sub-area is solved and processed to obtain the target restoration mode.

[0046] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0047] Acquire a plurality of pre-divided sub-areas of an active power distribution system;

[0048] For each sub-area, the switch action path is determined based on the pre-built Monte Carlo tree and the target power restoration model; the nodes of the Monte Carlo tree represent the topological state of the active distribution system, the switch action path represents the topological state of each stage of power restoration, and the target power restoration model is constructed based on the goal of minimizing the load value loss;

[0049] According to the switch action path and the target power supply restoration model, the power supply restoration mode of the sub-area is solved and processed to obtain the target restoration mode.

[0050] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0051] Acquire a plurality of pre-divided sub-areas of an active power distribution system;

[0052] For each sub-area, the switch action path is determined based on the pre-built Monte Carlo tree and the target power restoration model; the nodes of the Monte Carlo tree represent the topological state of the active distribution system, the switch action path represents the topological state of each stage of power restoration, and the target power restoration model is constructed based on the goal of minimizing the load value loss;

[0053] According to the switch action path and the target power supply restoration model, the power supply restoration mode of the sub-area is solved and processed to obtain the target restoration mode.

[0054] The above-mentioned method and device for determining the distributed power supply restoration mode of the distribution network obtains multiple sub-areas of the pre-divided active distribution system; for each sub-area, the switch action path representing the topological state of each stage of power supply restoration is determined based on the Monte Carlo tree and the target power supply restoration model constructed based on the minimum load value loss; according to the switch action path and the target power supply restoration model, the power supply restoration mode of the sub-area is solved and processed to obtain the target recovery mode. The Monte Carlo tree search of the present application effectively addresses the problem of the switch action sequence in the power supply restoration process by continuously exploring and evaluating different topological states, avoiding the computing power bottleneck faced by traditional methods when dealing with large-scale non-convex problems. The target power supply recovery model is used to solve the power supply recovery mode, and the target recovery mode obtained has good optimization performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 An application environment diagram of a method for determining a distributed power supply restoration mode of a power distribution network in one embodiment;

[0056] Figure 2 A schematic diagram of a flow chart of a method for determining a distributed power supply restoration mode of a power distribution network in one embodiment;

[0057] Figure 3 A schematic diagram of a flow chart for determining a switch action path in one embodiment;

[0058] Figure 4 A schematic diagram of a process for obtaining a leaf node to be processed in one embodiment;

[0059] Figure 5 A schematic diagram of a process for obtaining multiple sub-nodes in one embodiment;

[0060] Figure 6 A schematic diagram of a process for determining the value of a leaf node to be processed in one embodiment;

[0061] Figure 7 A schematic diagram of a process of constructing a target power supply restoration model in one embodiment;

[0062] Figure 8 A schematic diagram of a process for constructing an initial power restoration model in an embodiment;

[0063] Fig. 9 A schematic diagram of a process for constructing a target power supply restoration model in an embodiment;

[0064] Fig.10 A schematic diagram of a process of constructing a solution model in one embodiment;

[0065] Fig.11 A schematic diagram of an improved IEEE123 node network in an embodiment;

[0066] Fig.12 The IEEE123 node network partitioning situation in one embodiment;

[0067] Fig.13 The power supply recovery result of scheme I in one embodiment;

[0068] Fig.14 This is a power restoration result of Scheme II in an embodiment;

[0069] Fig.15 It is a structural block diagram of a device for determining a distributed power supply restoration mode of a power distribution network in one embodiment;

[0070] Fig.16 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0072] The method for determining the distributed power supply restoration mode of the distribution network provided in the embodiment of the present application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud, other network servers or in the terminal 102. The server obtains multiple sub-areas of the pre-divided active distribution system; for each sub-area, the switch action path representing the topological state of each stage of power supply recovery is determined based on the Monte Carlo tree and the target power supply recovery model constructed based on the minimum load value loss; according to the switch action path and the target power supply recovery model, the power supply recovery mode of the sub-area is solved and processed to obtain the target recovery mode. Among them, the terminal 102 can be the core control node of the entire active distribution network, receiving data information from sensors in each sub-area, including voltage, current, power, distributed power supply status, switch position, etc. It can also be a distributed power supply controller, an intelligent substation terminal, etc. The server 104 can be implemented with an independent server or a server cluster consisting of multiple servers.

[0073] In one embodiment, Figure 2 As shown, the present application provides a method for determining a distributed power supply restoration mode of a distribution network, and the method is applied to Figure 1 The server in the example is used to illustrate the following steps:

[0074] S201, obtaining a plurality of pre-divided sub-areas of an active power distribution system.

[0075] Among them, the active distribution system refers to a distribution system that includes distributed power sources (such as solar and wind power generation devices) and can actively control and manage the flow of electric energy. Compared with the traditional distribution system, it has stronger autonomy and flexibility, but the access of distributed power sources also increases the complexity of system operation.

[0076] In an embodiment of the present application, the nodes of the active power distribution network are first divided into two categories. Nodes connected to switches are defined as boundary nodes, which directly affect the electrical connection between regions when the switch is operated; nodes not connected to switches are called internal nodes. Then, sub-regions are formed by connecting boundary nodes and internal nodes with non-switch branches. On the premise of ensuring that no switches appear inside the sub-regions, the division scheme with the least number of sub-regions is selected as the sub-region division scheme of the active distribution system. For example, for an active distribution network containing multiple nodes and switches, all possible connection combinations can be traversed, the number of sub-regions under each combination can be calculated, and finally the optimal division method can be determined.

[0077] Another implementation method: sub-areas can be divided according to the distribution location of distributed power sources. Nodes that are close to distributed power sources and have close electrical connections are divided into a sub-area to fully utilize the local power supply capacity of distributed power sources. In specific operations, the location of the distributed power source is first determined, and then the sub-areas are gradually expanded to include surrounding nodes based on a certain electrical distance or connection relationship criterion, while ensuring that the connection between each sub-area meets the requirements of switch control and power supply recovery.

[0078] S202: For each sub-area, determine a switch action path based on a pre-built Monte Carlo tree and a target power supply restoration model.

[0079] Among them, the Monte Carlo tree is a search algorithm structure used in the decision-making process, and the nodes of the Monte Carlo tree represent the topological state of the active distribution system.

[0080] The meaning of the paths and nodes of the Monte Carlo tree is as follows: the paths in the Monte Carlo tree represent the paths of topological changes of the distribution system at various stages during the multi-node power supply restoration process; the nodes of the Monte Carlo tree represent the topological state of the distribution system, and each node includes n sw Group variables representing the distribution system topology

[0081] Among them, n sw Indicates the number of switches; is the variable representing the power flow direction of the switch branch, Indicates that the power flow direction of the switch branch is from node i to node j; is the variable representing the power flow direction of the switch branch, Indicates that the power flow direction of the switch branch is from node j to node i; is a 0 / 1 variable representing the state of the switch, Indicates that the switch is closed. Indicates the switch is on.

[0082] The nodes in the Monte Carlo tree are divided into root nodes, leaf nodes, and intermediate nodes; the root node represents a node without a parent node; the leaf node represents a node without a child node; the intermediate node represents a node with both a child node and a parent node; the connection between the child node and the parent node in the Monte Carlo tree represents the switch action plan, that is, the switch action plan that needs to be executed to transition from the distribution network topology state of the parent node to the distribution network topology state of the child node;

[0083] The value of a node in the Monte Carlo tree is expressed as: The value of a node in the Monte Carlo tree is the load value recovery amount of the distribution network under the topological state represented by the node, expressed as formula (1);

[0084]

[0085] Among them, v represents the value of node m in the Monte Carlo tree; represents the load loss rate of node i in the power supply restoration scheme under the topological state corresponding to node m in the Monte Carlo tree, and represents the proportion of the unrestored load of node i in the total load; Ω z represents the set of nodes in sub-region z; N z Represents the set of all sub-regions.

[0086] The switch action path refers to the path formed by the operation sequence of a series of switches during the power supply restoration process. The switch action at each stage will cause the topological state of the distribution system to change, and the switch action path records the topological change process at each stage from the initial fault state to the final complete restoration of power supply. The target power supply restoration model is built with the goal of minimizing the load value loss.

[0087] In the embodiment of the present application, linearized constraints can be used, including safe operation constraints (such as voltage and current limits, etc.), flow constraints (ensuring power balance, etc.), and distributed power operation constraints (limiting the output range of distributed power sources, etc.). When determining the switch action path, the topological state of each node is evaluated according to the Monte Carlo tree search process and the target power supply recovery model, and the path that can minimize the load value loss and meet the constraints is selected as the switch action path.

[0088] S203: performing a power supply restoration method solution process on the sub-area according to the switch action path and the target power supply restoration model to obtain a target restoration method.

[0089] Among them, the target recovery mode is calculated based on the switch action path and the target power supply recovery model. The specific operation mode of each sub-area in the power supply recovery process, including the determination of the load recovery amount and the arrangement of distributed power output, is used to achieve effective power supply recovery of the entire active distribution network.

[0090] In an embodiment of the present application, according to the determined switch action path, each sub-area performs switch operations in turn according to the change of the topological state in the path. At each stage, based on the target power supply recovery model, the power supply recovery method of the sub-area is determined by solving the variables in the model. Specifically, it includes calculating the load recovery amount and the output of distributed power sources. For example, using the flow constraints and distributed power operation constraint equations in the model, combined with the current topological state and system parameters, mathematical calculations are performed to derive the load amount that each node should restore in this state and the output size that the distributed power source should provide.

[0091] In the solution process, an iterative algorithm, such as the method of stepwise approximation, may be used to continuously adjust the value of the variable until the convergence condition of the model is met (such as the power balance error is within the allowable range, etc.). At the same time, each sub-area exchanges information with the adjacent area through limited communication, and coordinates to adjust their own recovery methods to achieve stable recovery of the entire active distribution network.

[0092] Another implementation method: A distributed collaborative optimization algorithm can be used to solve the power supply restoration method. The controller of each sub-area first preliminarily estimates the local power supply restoration plan based on local information and part of the information transmitted from the adjacent area. Then, the preliminary plan is adjusted and optimized through the information exchange and negotiation mechanism between regions. For example, information such as load demand and power supply capacity is exchanged between adjacent areas, and the local distributed power output and load recovery order are dynamically adjusted based on this information to achieve the optimal power supply restoration effect for the entire system. This method further improves the distributed collaborative capabilities of the system, reduces dependence on the central controller, and enhances the robustness of the system.

[0093] Alternatively, artificial intelligence algorithms (such as neural networks) can be used to learn historical power restoration data and build a prediction model for power restoration methods. In the actual power restoration process, the current switch action path and system status are used as inputs to quickly obtain a preliminary power restoration method through the prediction model, and then fine-tune it in combination with the target power restoration model. This can increase the solution speed, especially for power restoration scenarios with high real-time requirements.

[0094] The above-mentioned method and device for determining the distributed power supply restoration mode of the distribution network obtains multiple sub-areas of the pre-divided active distribution system; for each sub-area, the switch action path representing the topological state of each stage of power supply restoration is determined based on the Monte Carlo tree and the target power supply restoration model constructed based on the minimum load value loss; according to the switch action path and the target power supply restoration model, the power supply restoration mode of the sub-area is solved and processed to obtain the target recovery mode. The Monte Carlo tree search of the present application effectively addresses the problem of the switch action sequence in the power supply restoration process by continuously exploring and evaluating different topological states, avoiding the computing power bottleneck faced by traditional methods when dealing with large-scale non-convex problems. The target power supply recovery model is used to solve the power supply recovery mode, and the target recovery mode obtained has good optimization performance.

[0095] In an exemplary embodiment, based on the above embodiment, see Figure 3 The embodiment of the present application relates to a process of determining a switch action path based on a pre-built Monte Carlo tree and a target power supply recovery model, comprising the following steps:

[0096] S301, in each round of iteration, path exploration is performed based on the Monte Carlo tree to obtain leaf nodes to be processed.

[0097] Among them, the leaf node to be processed refers to the leaf node that needs to be further processed (such as simulation operation, value evaluation, etc.) in each round of iteration during the Monte Carlo tree search process. It represents a possible final or staged topological state on the current exploration path and is a key node representation of the current search direction. By processing it, more information about the path is obtained to decide whether to continue exploring along the path or turn to other paths.

[0098] In the embodiment of the present application, a priority-based search strategy can be adopted. In addition to the UCB value, a priority index is set for each child node, which comprehensively considers factors such as the stability of the node's topological structure (such as the number of connected nodes, whether a loop is formed, etc.), the proximity to the distributed power source, etc. When selecting the next exploration node, the child nodes are first sorted according to the priority, and then the child node with the largest UCB value is selected as the new current node among the higher priority child nodes. In this way, the exploration process can focus on more potential and more stable topological states more quickly, thereby improving search efficiency.

[0099] Alternatively, a random exploration factor is introduced. With a certain probability, instead of selecting the child node with the largest UCB value, a child node is randomly selected for exploration. This random exploration helps avoid falling into a local optimal solution, especially in a complex distribution network, where there may be multiple paths that appear to be optimal but are not actually globally optimal. Random exploration can increase the chance of discovering a better path.

[0100] S302, performing simulation operation processing on the leaf node to be processed, and determining the value of the leaf node to be processed according to the processing result and the target power supply recovery model.

[0101] Among them, the simulation operation is a calculation process based on the target power supply restoration model and related algorithms for the leaf nodes to be processed. By substituting the topological state represented by the leaf nodes to be processed into the target power supply restoration model and using a specific algorithm (such as a solution method based on the ADMM algorithm, etc.) for calculation, some key performance indicators or parameter estimates of the distribution network under this topological state, such as the load value recovery amount, are obtained, providing data support for the subsequent determination of the value of the leaf nodes to be processed.

[0102] In an embodiment of the present application, the topological state represented by the leaf node to be processed is substituted into the target power supply recovery model. In the simulation operation, the relevant parameters of the active distribution system obtained in step S201 (such as line parameters, load levels, topological connection relationships, etc. of each sub-area) and the topological state information of the current leaf node to be processed (such as switch status, etc.) are first taken as input. Then, calculations are performed based on the constraints and objective functions in the target power supply recovery model. For example, the power distribution under the topological state is calculated according to the power flow constraint equation, and the feasible output range of the distributed power source is determined according to the distributed power source operation constraints. The model is solved using methods such as the ADMM algorithm to obtain the load value recovery amount of the distribution network under the topological state. Based on the load value recovery amount, the value of the leaf node to be processed is calculated using a formula.

[0103] S303, updating a Monte Carlo tree according to the value of the leaf node to be processed, terminating the iteration when the number of iterations meets a preset condition, and determining the search path of the Monte Carlo tree as the switch action path.

[0104] In an embodiment of the present application, after obtaining the value of the leaf node to be processed, the Monte Carlo tree is updated. It is determined whether the value of the leaf node to be processed is less than the value of its parent node. If it is less, the leaf node to be processed is deleted from the Monte Carlo tree, and the number of explorations of its parent node is increased by 1; if it is greater, the leaf node to be processed is retained, and the number of explorations of it and its parent node are increased by 1. In this way, the structure of the Monte Carlo tree is constantly adjusted, and gradually tends to retain paths with higher values. Repeat the process of S301-S303 for the next round of iterations. When the number of iterations reaches a preset condition (such as a preset maximum number of iterations), the iteration is stopped. At this time, the path from the root node to the last retained leaf node in the Monte Carlo tree is the search path, which is determined as the switch action path. The switch action path records the order of changes in the topological state of the distribution system at each stage during the power supply recovery process, that is, the operation sequence of a series of switches.

[0105] In an exemplary embodiment, based on the above embodiment, see Figure 4 The embodiment of the present application relates to a process of performing path exploration processing based on a Monte Carlo tree to obtain a leaf node to be processed, including the following steps:

[0106] S401, for any node in the Monte Carlo tree, when the node has not been subjected to path exploration processing, an expansion operation is performed on the node to obtain multiple child nodes.

[0107] In an embodiment of the present application, first, determine the current topological state of the power distribution system represented by the node, including the current state of each switch and the connection relationship between the nodes. Then, operations are performed on the switch state, and the switch actions of the power distribution system are defined to include closing a switch, opening a switch, closing a switch and then immediately opening a switch, and other operation types. For each possible switch action, a new topological state is obtained after executing the operation, which is used as a child node of the node. For example, if there are n switches in the topological state represented by the current node, the above three operations are tried for each switch (certain constraints must be met, such as not violating safe operation constraints, etc.), then a maximum of 3n child nodes can be obtained. In the process of generating child nodes, it is necessary to check the legitimacy of the new topological state to ensure that the basic operating rules of the power distribution system are met, such as not forming a short circuit and not exceeding the line capacity.

[0108] S402, calculating the upper confidence limit value of each child node. The upper confidence limit value is used to evaluate the estimated value of each child node.

[0109] S403: Determine the child node with the maximum upper confidence limit value as the leaf node to be processed.

[0110] The specific meaning of path search is: finding a path from the root node to the leaf node in the Monte Carlo tree, that is, finding a solution to achieve the transition of the distribution system from the initial topological state to the final topological state through switching actions.

[0111] In the embodiment of the present application, the specific steps of performing path search starting from the root node of the Monte Carlo tree are:

[0112] (1) Take the root node of the Monte Carlo tree as the current node;

[0113] (2) Use formula (2) to calculate the upper confidence bounds of all child nodes of the current node;

[0114]

[0115] Among them, UCB m represents the upper confidence bound of node m, represents the average value of the current node and its child nodes; c is the equilibrium constant; N represents the total number of iterations; n i Indicates the number of explorations of the node;

[0116] (3) Select the child node with the largest upper confidence bound as the new current node;

[0117] (4) Determine whether the current node is a leaf node. If it is not a leaf node, repeat (2) and (3). If it is, end the search and use the current node as the leaf node to be processed.

[0118] The specific meaning of the pending leaf node is: the pending leaf node represents the node that needs to perform operations such as expansion, simulation, and update of the Monte Carlo tree in this round of iteration. There can only be one pending leaf node in the entire Monte Carlo tree.

[0119] The embodiment of the present application determines the child node with the maximum upper confidence bound value as the leaf node to be processed, providing a clear direction for path exploration. By selecting the child node with the largest UCB value, it is possible to continue to explore in depth along the most potential path and gradually approach the optimal power restoration path. Improvements such as multi-criteria decision-making methods and random selection mechanisms further optimize the selection process of leaf nodes to be processed, improve the comprehensiveness and flexibility of path exploration, and enhance the algorithm's ability to find the optimal solution in a complex active distribution network environment, thereby improving the quality and reliability of the power restoration strategy.

[0120] In an exemplary embodiment, based on the above embodiment, see Figure 5 The embodiment of the present application relates to a process of performing an expansion operation on a node to obtain multiple child nodes, including the following steps:

[0121] S501, enumerate the topological states of the nodes to obtain multiple initial switch action plans.

[0122] In the embodiment of the present application, first, the current topological state of the node is clarified, including the current open and close positions of each switch and the electrical connection relationship between the nodes. Then, for each switch, its possible action is considered, that is, closed or open. For a node with n switches, each switch has two possible states (open or closed), then the total number of initial switch action schemes is at most 2 n (considering the case where all switches are actuated simultaneously). However, in practical applications, some obviously unreasonable combinations may be excluded based on some prior knowledge or rules, such as disconnecting all switches connected to the power supply at the same time. For example, in a simple node topology containing 3 switches, the possible initial switch action schemes are: switch 1 closed, switch 2 closed, switch 3 closed; switch 1 closed, switch 2 closed, switch 3 opened; switch 1 closed, switch 2 opened, switch 3 closed; switch 1 closed, switch 2 opened, switch 3 opened; switch 1 opened, switch 2 closed, switch 3 closed; switch 1 opened, switch 2 closed, switch 3 closed; switch 1 opened, switch 2 closed, switch 3 closed, switch 1 opened, switch 2 closed, switch 3 opened, switch 1 opened, switch 2 closed, switch 3 closed, switch 3 opened, switch 1 opened, switch 2 closed, switch 3 opened, switch 3 opened, switch 3 opened, switch 1 opened, switch 2 opened, switch 3 closed; switch 1 opened, switch 2 opened, switch 3 opened, a total of 8 schemes (without considering other restrictions).

[0123] S502, determining a plurality of feasible switching action plans according to each initial switching action plan and predetermined state constraints.

[0124] In the embodiment of the present application, for each initial switch action scheme, it is substituted into the predetermined state constraint conditions for verification. The state constraint conditions include power flow constraints, that is, the power flow in the power grid under the switch action scheme is calculated according to Kirchhoff's law to ensure the power balance of each node (the inflow power is equal to the outflow power), and the power of each branch does not exceed its line capacity limit; in terms of safe operation constraints, check whether the voltage of each node is within the allowable upper and lower limits, and whether the maximum square difference of the voltage allowed at both ends when the switch is disconnected meets the requirements; for distributed power supply operation constraints, verify whether the active and reactive output of the distributed power supply is within its set capacity, active and reactive output limits. For example, for an initial switch action scheme, after calculating the resulting power flow distribution, it is found that the power of a certain line exceeds its capacity limit, then the scheme does not meet the requirements and is determined to be infeasible. Only initial switch action schemes that meet all constraints are determined as feasible switch action schemes.

[0125] S503, executing each feasible switch action scheme on the node respectively, and determining the topological state after each feasible switch action scheme is executed as a child node.

[0126] In an embodiment of the present application, for each determined feasible switch action scheme, a corresponding switch action is performed based on the current topological state of the node. For example, if a feasible scheme is to close a switch, then the state of the switch is changed from open to closed in the current topological state to obtain a new topological state. This new topological state is the result after executing the feasible switch action scheme, and it is determined as a child node of the node. For multiple feasible switch action schemes, this process is repeated to obtain multiple child nodes, each of which represents a possible topological state of the distribution system after executing the corresponding feasible switch action scheme. These child nodes will constitute the next layer of nodes of the node for subsequent operations such as path exploration and value assessment.

[0127] In an embodiment of the present application, it is determined whether the leaf node to be processed in the Monte Carlo tree has been explored. If it has been explored, an expansion operation is performed on the leaf node to be processed to obtain the child nodes of the node, and the first child node obtained is set as the leaf node to be processed, and a simulation operation is performed on it; if it has not been explored, a simulation operation is performed on the leaf node to be processed.

[0128] In the embodiment of the present application, the specific steps of the expansion operation are:

[0129] (1) The definition of switching actions in the power distribution system includes: closing a switch, opening a switch, and closing a switch and then immediately opening a switch;

[0130] (2) The controller of each region exhaustively searches for all the switch action plans of the region under the topological state of the leaf node to be processed in the Monte Carlo tree, and finds the switch action plan that satisfies constraint (3), which is recorded as a feasible switch action plan;

[0131] (3) The topological state of the distribution network after the execution of each feasible switch action plan is taken as a child node of the leaf node to be processed in the Monte Carlo tree.

[0132]

[0133] in, is a variable representing the power flow direction of the switch branch in stage n (the power supply restoration stage represented by the leaf node to be processed), Indicates that the power flow direction of the switch branch in stage n is from node i to node j; is the variable representing the power flow direction of the switch branch in stage n, Indicates that the power flow direction of the switch branch in stage n is from node j to node i; is a 0 / 1 variable representing the state of the switch in stage n, Indicates that the switch is closed in stage n, Indicates that the switch is disconnected in stage n; s z,n Indicates the power-on state of region z in stage n; s z,n-1 Indicates the power-on status of area z in stage n-1 (power restoration stage represented by the parent node); s z′,n-1 Indicates the power-on state of the region z' adjacent to the region z in stage n-1; Ω i,s represents the set of boundary nodes connected to node i; Represents the set of sub-areas equipped with distributed generation.

[0134] In an embodiment of the present application, the topological state after the execution of a feasible switch action plan is determined as a child node, which provides a practical expansion direction for the construction and path exploration of the Monte Carlo tree. By generating child nodes, a tree structure describing various possible paths in the power supply restoration process is gradually constructed, so that the algorithm can perform effective search and decision-making on this structure. Considering the sequence of switch actions and time delays, as well as preprocessing operations on child nodes, the accuracy and information richness of child nodes are further improved, which helps to more accurately evaluate the pros and cons of the path, thereby selecting a better power supply restoration strategy and improving the quality and efficiency of power supply restoration in the active distribution network.

[0135] In an exemplary embodiment, based on the above embodiment, see Figure 6 The embodiment of the present application relates to a process of performing a simulated operation on a leaf node to be processed, and determining the value of the leaf node to be processed according to the processing result and the target power supply recovery model, including the following steps:

[0136] S601, obtaining initial parameters of the leaf node to be processed.

[0137] In the embodiment of the present application, all parameter information related to the leaf node to be processed is extracted from the database or real-time monitoring system of the active distribution network. First, according to the position of the leaf node to be processed in the Monte Carlo tree and the topological state identifier represented, the corresponding physical node and connection relationship are determined. Then, the line parameters of these nodes are obtained, including the resistance and reactance values ​​of each line, which determine the transmission loss and characteristics of electric energy in the line. Next, the load level of each node is obtained, and the specific values ​​of active load and reactive load are recorded respectively, which reflects the power demand of the node. At the same time, the topological connection relationship data of the distribution network is obtained to clarify the connection mode between nodes, such as which nodes are connected through switches and which are directly connected. It is also necessary to obtain the current voltage value of the node for evaluating the power quality and operating status of the node. In addition, the state of each switch (on or off) is determined, which is one of the key information of the topological state. For nodes with distributed power sources, relevant parameters are obtained, including the type of distributed power source (such as solar energy, wind energy, etc.), access location (connected to which node), capacity (maximum output power) and current active and reactive output values. All these parameters are organized into a format suitable for input into the target power restoration model, such as forming a parameter vector or matrix form.

[0138] S602, inputting the initial parameters into the target power supply recovery model to solve the power supply problem and obtain the load value recovery amount of the leaf node to be processed.

[0139] In the embodiment of the present application, the initial parameters of the leaf nodes to be processed are obtained and substituted into the target power supply recovery model. The target power supply recovery model is calculated according to its built-in objective function (minimization of load value loss) and constraints (safe operation, power flow, distributed power supply operation and other constraints). First, according to the power flow constraint equation, the power distribution in the topological state is calculated using information such as the voltage, line parameters and load level of the node to ensure the power balance of each node (that is, the inflow power is equal to the outflow power). At the same time, according to the safe operation constraint, check whether the node voltage is within the allowable upper and lower limits, and whether the current in the line exceeds its capacity limit. If not, make corresponding adjustments or mark it as an infeasible solution. For the distributed power supply operation constraints, verify whether its current output is within the set capacity and limit range. If it exceeds the range, limit its output or optimize and adjust it. Under the premise of meeting all constraints, the objective function is solved by an optimization algorithm (such as a linear programming algorithm, an interior point method, etc.), and the load value that can be restored in the topological state, that is, the load value recovery amount, is calculated. For example, by calculating the recovery status of different load nodes and combining the load importance weight value of each node, the total load value recovery amount can be obtained.

[0140] S603, inputting the load value recovery amount of the leaf node to be processed into a pre-built value calculation model to perform load value calculation processing to obtain the value of the leaf node to be processed.

[0141] The specific steps are:

[0142] (1) The rough convergence threshold is used as the convergence threshold of the ADMM algorithm. The controllers in each area use the established method for solving the active distribution network power supply restoration problem based on the ADMM algorithm to obtain the distribution network load value recovery amount under the topological state of the current node in the Monte Carlo tree;

[0143] (2) According to the distribution network load value recovery amount under the topological state of the current node in the Monte Carlo tree, the value of the leaf node to be processed is calculated using formula (1).

[0144] Determine whether the leaf node to be processed meets the path feasibility constraint (4), that is, whether the value of the leaf node to be processed is less than the value of the parent node. If so, the exploration times of all parent nodes of the leaf node to be processed is increased by 1, and the leaf node to be processed is deleted from the Monte Carlo tree; if so, the exploration times of the leaf node to be processed and its parent nodes is increased by 1;

[0145] v n ≥v n，p (4)

[0146] Among them, v n Indicates the value of the leaf node to be processed, v n,p Indicates the value of the parent node of the leaf node to be processed.

[0147] The embodiment of the present application converts the load value recovery amount into the value of the leaf node to be processed through the value calculation model, and provides a unified evaluation standard for the nodes in the Monte Carlo tree. This value can comprehensively reflect the various performances of the node in the power supply restoration process, which is convenient for comparison and selection in the tree structure. The adaptive weight adjustment mechanism makes the value calculation model more flexible, and can adjust the evaluation focus according to actual needs, thereby improving the practicality of the evaluation results. Using machine learning technology to build a model can better mine the hidden relationships in the data and improve the accuracy of value calculation, especially when dealing with complex nonlinear relationships, and provide a more scientific decision-making basis for selecting the optimal power supply restoration path.

[0148] In an exemplary embodiment, based on the above embodiment, see Figure 7 The process of constructing the target power supply restoration model in the embodiment of the present application includes the following steps:

[0149] S701: construct an initial power supply restoration model based on pre-acquired power distribution system parameter information.

[0150] Among them, the initial power supply restoration model is a model framework initially established based on the distribution system parameter information, which mainly includes the basic description of the power supply restoration problem and some preliminary constraint setting. It may have considered some operating rules of the distribution system, such as simple power balance relationships, but has not yet been optimized and improved. It is the starting point for building the target power supply restoration model.

[0151] In the embodiment of the present application, the pre-acquired distribution system parameter information is first sorted and analyzed. The connection status of nodes and branches is determined according to the topological connection relationship, and mathematical models such as node-branch association matrix or adjacency matrix are constructed to describe the topological structure of the distribution network. For line parameters, they are associated with corresponding branches so that the influence of line resistance and reactance on power transmission can be considered in subsequent calculations. According to the load level, the power demand of each node is determined and used as the load parameter in the model. Then, according to the principle of power balance, the power flow equation is preliminarily established, that is, the injected power of each node is equal to the outflow power, considering the balance relationship between active power and reactive power.

[0152] S702: construct a target power supply restoration model according to the initial power supply restoration model and the solution model.

[0153] In the embodiment of the present application, the characteristics and requirements of the initial power supply restoration model are analyzed, and a suitable solution model is selected to be combined with it. If the constraints and objective functions in the initial model are mostly linear relationships, a linear programming solution model, such as the simplex method, can be selected. The equations and constraints in the initial power supply restoration model are converted into a standard form suitable for the selected solution model. For example, for the linear programming solution model, the objective function and the constraints are expressed in the form of linear inequalities or equations. Then, the converted initial power supply restoration model is embedded in the framework of the solution model, and the corresponding relationship between the variables in the solution model and the parameters in the initial model, as well as the mapping relationship between the objective function and the constraints, is determined. In this process, some adjustments and optimizations may be required to the initial model, such as adding slack variables or residual variables to handle inequality constraints, to ensure that the solution model can correctly handle various situations in the initial model. The initial power supply restoration model is optimized and calculated by the solution model to obtain a solution that meets certain optimization goals (such as minimum load value loss), and finally the target power supply restoration model is constructed. For example, during the solution process, the solution model will adjust the values ​​of variables (such as the output of distributed generation, the state of switches, etc.) according to the set optimization objectives to find a power restoration plan that minimizes the loss of load value while satisfying all constraints.

[0154] In an exemplary embodiment, based on the above embodiment, see Figure 8The power distribution system parameter information of the embodiment of the present application includes at least one of the multiple nodes in the sub-area, the total active load of each node, the total reactive load, and the load importance weight value; based on the pre-acquired power distribution system parameter information, constructing an initial power supply restoration model includes the following steps:

[0155] S801, determining an objective function and a first constraint condition according to the distribution system parameter information; the first constraint condition includes a safety operation constraint, a power flow constraint, and a distributed power generation operation constraint; the objective function is based on the total active load, the load importance weight value, and the load loss rate; the objective function is used to characterize the priority of different loads in power supply restoration;

[0156] S802, forming an initial power supply restoration model by means of the objective function and the first constraint condition.

[0157] In the embodiment of the present application, the minimum load value loss is set as the objective function, and linearized safety operation constraints, power flow constraints, and distributed power supply operation constraints are adopted, wherein:

[0158] The objective function of setting the load value loss to be minimum is expressed as formula (5):

[0159]

[0160] in, represents the total amount of active load of node i; e i represents the load importance weight value of node i, l i is the load loss rate of node i;

[0161] The linearized power flow constraint, safe operation constraint, and the first constraint condition of distributed power generation operation are expressed as formula (6):

[0162]

[0163]

[0164] Among them, Ω i，p represents the set of parent nodes of node i; P hi represents the active power flowing from node h to node i; represents the active power injected by the distributed generation of node i; Ω i，c represents the set of child nodes of node i; P ij represents the active power flowing from node i to node j; represents the active power of the load at node i; Q hi represents the reactive power flowing from node h to node i; represents the reactive power injected by the distributed generation at node i; Q ijrepresents the reactive power flowing from node i to node j; Represents the reactive power of the load at node i; U i represents the square of the voltage at node i; U j represents the square of the voltage at node j; r ij represents the impedance of line ij; x ij Represents the impedance of line ij; Ω in represents a node that is not connected to a switch; represents the square of the voltage of node j outside the sub-region connected to the switch; Ω sw represents the node connected to the switch; Ω z′ represents the set of regions adjacent to region z; Indicates the maximum square difference of voltage allowed across the two ends when the switch is disconnected; V i min Represents the minimum voltage allowed at node i; V i max represents the maximum voltage allowed for node i; M represents a very large number, which is 10000; S ij is an auxiliary variable; represents the capacity of the line between nodes i and j; Indicates the active power limit of distributed generation; Indicates the reactive power limit of distributed generation.

[0165] The initial power supply restoration model constructed by the embodiment of the present application can accurately reflect the power supply restoration requirements and operating characteristics of the distribution system by reasonably determining the objective function and constraints. The objective function is based on the total active load, the load importance weight value and the load loss rate, which effectively reflects the priority of different loads in power supply restoration, helps to prioritize the restoration of important loads under limited resources, and improves the efficiency of power supply restoration. The safe operation constraint ensures the safety of the system during the power supply restoration process, avoids equipment damage and safety accidents. The flow constraint ensures the reasonable flow of electric energy in the network and maintains the stable operation of the system. The distributed power supply operation constraint enables the distributed power supply to provide support for power supply restoration within a safe and stable range. The improved objective function and constraints (such as the introduction of time factors, dynamic adjustment, etc.) further improve the flexibility, adaptability and accuracy of the model, so that it can better cope with the complex situation of the actual distribution system during the power supply restoration process, optimize the power supply restoration strategy, and improve the efficiency and reliability of power supply restoration.

[0166] In an exemplary embodiment, based on the above embodiment, see Fig. 9 The embodiment of the present application relates to a process of constructing a target power supply restoration model according to an initial power supply restoration model and a solution model, comprising the following steps:

[0167] S901, based on the node topology state of the Monte Carlo tree, perform variable assignment processing on the initial power supply restoration model to obtain the assigned initial power supply restoration model.

[0168] In the embodiment of the present application, first, the topological state information of the current node in the Monte Carlo tree is analyzed. The connection relationship between the nodes is determined, that is, which nodes are directly connected, which nodes are connected through switches, and the current state of the switches (closed or open). For variables representing the node connection relationship (such as adjacency matrix elements, etc.), they are assigned according to the topological state.

[0169] Next, the voltage information of the node is obtained. According to the monitoring system or the preset initial conditions, the current voltage value of each node is determined and assigned to the voltage variable in the initial power supply restoration model.

[0170] Then, determine the power situation in the line. According to the voltage, connection relationship and load situation of the node, calculate or measure the active power and reactive power on each line, and assign them to the corresponding power variables in the model. For the lines connecting nodes, the active power and reactive power are obtained by the power flow calculation method (such as calculation based on Kirchhoff's law), and then the active power and reactive power are respectively assigned to the line power variables in the model.

[0171] For distributed power sources, obtain their current operating parameters, including the type of distributed power sources, access location, current active output and reactive output, etc. Based on this information, assign the active output value of the distributed power source to the distributed power source active output variable in the model, and assign the reactive output value to the reactive output variable.

[0172] Finally, consider the load situation. Obtain the load information on each node, including the total active load and the total reactive load, and assign them to the load variables in the model.

[0173] S902, forming a target power supply restoration model by using the assigned initial power supply restoration model and the solution model.

[0174] In an embodiment of the present application, the assigned initial power supply recovery model is embedded into the framework of the solution model. The corresponding relationship between the variables in the solution model and the variables in the assigned initial power supply recovery model is determined to ensure that the two can interact correctly. For example, a mapping relationship is established between the decision variables in the solution model (such as switch action variables, distributed power output adjustment variables, etc.) and the corresponding variables in the initial power supply recovery model, so that the solution model can operate and optimize the variables in the initial power supply recovery model.

[0175] The embodiment of the present application is composed of a target power supply restoration model consisting of an assigned initial power supply restoration model and a solution model, which can give full play to the advantages of both. The solution model provides the initial power supply restoration model with the ability to optimize the solution, so that it can find the optimal or approximately optimal power supply restoration plan under complex constraints. The adaptive solution model combination method improves the adaptability of the model to different situations, ensuring efficient solutions under various distribution system characteristics and scenarios. Establishing a model iteration update mechanism enables the model to have self-optimization capabilities, and can be continuously improved according to actual conditions, thereby improving the quality and reliability of the power supply restoration strategy and better meeting the actual needs of the active distribution network in the power supply restoration process.

[0176] In an exemplary embodiment, based on the above embodiment, see Fig.10 , the process of constructing the solution model in the embodiment of the present application includes:

[0177] S1001, decomposing the target function to obtain multiple sub-functions.

[0178] Among them, the sub-function is composed of iteration variables and dual variables; based on the convergence parameters obtained in advance, the convergence conditions are constructed; the convergence parameters include the current boundary variables and the first K boundary variables of the iteration variables;

[0179] S1002, forming a solution model by using the pre-acquired second constraint condition, each sub-function and convergence condition.

[0180] In the embodiment of the present application, the variables in the active distribution network power supply restoration model established are adjusted according to the topological state of the current node in the Monte Carlo tree. Assign values, and decompose the active distribution network power supply restoration problem after assignment into sub-problem 1 and sub-problem 2, where:

[0181] The sub-function of sub-problem 1 is expressed as formula (7), and the constraint condition is the second constraint condition of the established active distribution network power supply restoration model:

[0182]

[0183] in, represents the active power flowing from node i to node j in the internal branch of the sub-area; represents the reactive power flowing from node i to node j in the internal branch of the sub-area; represents the square of the voltage at the boundary node; σ k Indicates the size of the penalty factor at the kth iteration; N represents the variable of subproblem 2 obtained in the k-1th iteration; sw represents the set of boundary nodes in sub-region z, N pRepresents the set of parent nodes of node i, N c Represents the set of child nodes of node i; represents the dual variable obtained in the k-1th iteration;

[0184] The sub-function of sub-problem 2 is expressed as formula (8), and the constraints are the constraints of the established active distribution network power supply restoration model:

[0185]

[0186] in, Represents the variable obtained at the kth iteration The value of represents the variables of sub-problem 2;

[0187] (2) According to formula (9), solve sub-problem 1:

[0188] X *，k =argminf I (X, Y *，k-1 , Λ *，k-1 ) (9)

[0189] Among them, variable X represents the variables of sub-problem 1, including U i , l i , X br Represents the sub-region internal variables in variable X, including U i , l i ; represents the active power flowing from node i to node j in the internal branch of the sub-area; represents the reactive power flowing from node i to node j in the internal branch of the sub-area; X sw Represents the boundary variables in variable X, including The variable Y represents the variables of sub-question 2, including The variable Λ represents the dual variable, including μ ij , κ ij , π i , χ ij , ij , σ i ;X *，k represents the value of variable X solved at the kth iteration; Y *，k-1 represents the value of Y solved in the k-1th iteration, Λ *，k-1 represents the value of the dual variable Λ obtained in the k-1th iteration;

[0190] (3) According to formula (10), solve sub-problem 2:

[0191] Y *，k =argminf II (X *，k , Y, Λ *，k-1 ) (10)

[0192] Among them, Y *，k Represents the value of variable Y solved at the kth iteration;

[0193] (4) According to formula (11), update the dual variables:

[0194]

[0195] Among them, Λ *，k represents the value of the dual variable Λ solved at the kth iteration; represents the value of the boundary variable in the variable X solved at the kth iteration;

[0196] (5) According to formula (12), the penalty factor is updated:

[0197]

[0198] Among them, σ k+1 represents the size of the penalty factor at the k+1th iteration; h represents the penalty factor growth coefficient; ρ max Indicates the maximum value of the penalty factor;

[0199] (6) According to formula (13), judge the convergence:

[0200]

[0201] Where i represents the convergence threshold of the ADMM algorithm, Represents the residual of distributed computing. When formula (10) is satisfied, the ADMM algorithm converges. Otherwise, it returns to (2) and continues iterative computing until convergence.

[0202] A specific example is given below to verify the effect of this application.

[0203] First, input the impedance value of the line element in the improved 94-node system, the active power reference value and power factor of the load element, and the network topology connection relationship. The example structure is as follows: Fig.11 As shown, detailed parameters are shown in Table 1, Table 2, Table 3 and Table 4; due to extreme faults, transformers 1, 17, 28, and 34 failed. The maximum active output of all distributed power sources is 300kW, and the maximum reactive output is 200kva rThe system's reference voltage is set to 11.4 kV, the reference power is set to 1 MVA, and each value in the system is normalized; finally, the upper and lower limits of the voltage amplitude (normalized value) of each node are set to be 1.05 and 0.95 respectively; the initial value of the penalty factor ρ is 0.1, the penalty factor growth coefficient is 1.2, the maximum value of the penalty factor is 4, the rough convergence standard ε of the ADMM algorithm is 0.01, and the precise convergence standard ε of the ADMM algorithm is 0.001; the balance coefficient of the Monte Carlo tree search algorithm is 200.

[0204] Table 1 Load access location and power of the improved 94-node example

[0205]

[0206]

[0207] Table 2 Improved load importance weight values ​​of 94 nodes

[0208]

[0209]

[0210] Table 3 Improved 94-node line parameters

[0211]

[0212] Table 4 Improved switch branch parameters of 94-node example

[0213]

[0214]

[0215] Two schemes are used for comparative analysis:

[0216] Solution I: Use the distribution network distributed power supply restoration mode confirmation method of this application for processing.

[0217] Solution II: Use the ADMM algorithm to solve the single time section power supply restoration problem.

[0218] The computer hardware environment for performing optimization calculations is Intel(R) Core(TM) i7-10750H, with a main frequency of 2.59GHz and a memory of 16GB; the software environment is the Windows 11 operating system.

[0219] The improved 94-node network partitioning result of the distribution network distributed power supply restoration mode confirmation method of the present application is as follows: Fig.12 The power supply restoration strategy solved by this method is shown in Table 5, and the final power supply restoration result is shown in Fig.13 Scheme II is used to verify the difference between the power supply restoration strategy generation speed of Scheme I and other distributed methods. The final power supply restoration result of Scheme II is shown in Fig.14 shown.

[0220] Table 5 Power restoration strategy generated by Scheme I

[0221] stage Switching action Load value recovery Number of areas where power was restored Phase 1 Close switch 10 9982.8 9 Phase 2 Close switch 4 11801.5 10 Phase 3 Close switch 1 12521.1 11 Phase 4 Close switch 13 12699.0 11

[0222] Table 6 Comparison of calculation time between method I and method II

[0223] method Iterations The time for each iteration Calculation time Method I 278 0.12 33.36s Method II 200 0.27 54.00s

[0224] After adopting the power supply restoration method of the present application, after an extreme fault occurs, each area only needs to communicate with the adjacent area to achieve distributed power supply restoration. This method reduces the computational burden under acceptable optimization effect. Taking a 94-node network as an example, the computational time required for method I and method II is shown in Table 6. Compared with using the ADMM algorithm to solve the single time section power supply restoration problem, this method can effectively reduce the controller's computational burden and reduce the computational time.

[0225] Table 6 Comparison of calculation time between method I and method II

[0226] Way Iterations The time for each iteration Calculation time Method I 278 0.12 33.36s Method II 200 0.27 54.00s

[0227] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0228] Based on the same inventive concept, the embodiment of the present application also provides a distribution network distributed power supply recovery mode determination device for implementing the above-mentioned distribution network distributed power supply recovery mode determination method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in one or more distribution network distributed power supply recovery mode determination device embodiments provided below can refer to the above-mentioned limitations on the distribution network distributed power supply recovery mode determination method, and will not be repeated here.

[0229] In one embodiment, Fig.15 As shown, a device 1100 for determining a distributed power supply restoration mode of a distribution network is provided, comprising:

[0230] The area acquisition module 1101 is used to acquire a plurality of sub-areas of the pre-divided active power distribution system;

[0231] The path determination module 1102 is used to determine the switch action path for each sub-area based on the pre-built Monte Carlo tree and the target power supply restoration model; the nodes of the Monte Carlo tree represent the topological state of the active power distribution system, the switch action path represents the topological state of each stage of power supply restoration, and the target power supply restoration model is constructed based on the goal of minimizing the load value loss;

[0232] The solving module 1103 is used to solve the power supply restoration mode of the sub-area according to the switch action path and the target power supply restoration model to obtain the target restoration mode.

[0233] In one embodiment, the path determination module includes:

[0234] A path exploration unit is used to perform path exploration based on the Monte Carlo tree in each round of iteration to obtain leaf nodes to be processed;

[0235] A simulation operation unit, used for performing simulation operation processing on the leaf node to be processed, and determining the value of the leaf node to be processed according to the processing result and the target power supply recovery model;

[0236] The path determination unit is used to update the Monte Carlo tree according to the value of the leaf node to be processed, terminate the iteration when the number of iterations meets the preset conditions, and determine the search path of the Monte Carlo tree as the switch action path.

[0237] In one embodiment, the path exploration unit includes:

[0238] The child node determination subunit is used to perform an expansion operation on any node in the Monte Carlo tree to obtain multiple child nodes when the node has not been subjected to path exploration.

[0239] The estimated value determination subunit is used to calculate the upper confidence limit value of each child node; the upper confidence limit value is used to evaluate the estimated value of each child node;

[0240] The processing node determination subunit is used to determine the child node with the maximum upper confidence limit value as the leaf node to be processed.

[0241] In one embodiment, the subnode determination subunit is specifically used for:

[0242] The topological states of the nodes are enumerated to obtain multiple initial switch action plans;

[0243] Determine multiple feasible switching action plans according to each initial switching action plan and predetermined state constraints;

[0244] Each feasible switch action plan is executed on the node respectively, and the topological state after each feasible switch action plan is executed is determined as a child node.

[0245] In one embodiment, the simulation operation unit includes:

[0246] A parameter acquisition subunit, used to acquire the initial parameters of the leaf node to be processed;

[0247] A solving subunit is used to input the initial parameters into the target power supply recovery model to solve the power supply problem and obtain the load value recovery amount of the leaf node to be processed;

[0248] The value calculation subunit is used to input the load value recovery amount of the leaf node to be processed into a pre-built value calculation model to perform load value calculation processing to obtain the value of the leaf node to be processed.

[0249] In one embodiment, the above device further comprises:

[0250] An initial model building module, used to build an initial power supply restoration model based on the pre-acquired distribution system parameter information;

[0251] The target model building module is used to build a target power supply restoration model according to the initial power supply restoration model and the solution model.

[0252] In one embodiment, the power distribution system parameter information includes at least one of a plurality of nodes in the sub-area, a total active load of each node, a total reactive load, and a load importance weight value; and the initial model building module includes:

[0253] A parameter determination unit is used to determine an objective function and a first constraint condition according to the distribution system parameter information; the first constraint condition includes a safety operation constraint, a power flow constraint and a distributed power supply operation constraint; the objective function is based on the total active load, the load importance weight value and the load loss rate; the objective function is used to characterize the priority of different loads in power supply restoration;

[0254] The initial model building unit is used to form an initial power supply restoration model based on the objective function and the first constraint condition.

[0255] In one embodiment, the target model building module includes:

[0256] An initial model determination unit is used to perform variable assignment processing on the initial power supply restoration model based on the node topology state of the Monte Carlo tree to obtain the assigned initial power supply restoration model;

[0257] The target model building unit is used to form a target power supply restoration model from the assigned initial power supply restoration model and the solution model.

[0258] In one embodiment, the above device further comprises:

[0259] Function decomposition module, used to decompose the target function to obtain multiple sub-functions; the sub-functions are composed of iteration variables and dual variables;

[0260] A convergence condition determination module is used to construct a convergence condition based on pre-acquired convergence parameters; the convergence parameters include the current boundary variables and the previous K boundary variables of the iteration variables;

[0261] The solution model determination module is used to form a solution model from the pre-acquired second constraint conditions, each sub-function and convergence condition.

[0262] Each module in the above-mentioned distribution network distributed power supply restoration mode determination device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each of the above modules.

[0263] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Fig.16 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for determining a distributed power supply recovery mode of a distribution network is implemented.

[0264] Those skilled in the art will understand that Fig.16The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0265] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0266] Acquire a plurality of pre-divided sub-areas of an active power distribution system;

[0267] For each sub-area, the switch action path is determined based on the pre-built Monte Carlo tree and the target power restoration model; the nodes of the Monte Carlo tree represent the topological state of the active distribution system, the switch action path represents the topological state of each stage of power restoration, and the target power restoration model is constructed based on the goal of minimizing the load value loss;

[0268] According to the switch action path and the target power supply restoration model, the power supply restoration mode of the sub-area is solved and processed to obtain the target restoration mode.

[0269] According to some embodiments of the present application, a computer program product is also provided, and when the computer program is executed by a processor, the above method can be implemented. The computer program product includes one or more computer instructions. When these computer instructions are loaded and executed on a computer, part or all of the above method can be implemented in whole or in part according to the process or function described in the embodiment of the present application.

[0270] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0271] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0272] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0273] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for determining a distributed power supply restoration mode of a distribution network, characterized in that: The method comprises: Acquire a plurality of pre-divided sub-areas of an active power distribution system; For each of the sub-areas, a switch action path is determined based on a pre-constructed Monte Carlo tree and a target power supply restoration model; the nodes of the Monte Carlo tree represent the topological state of the active power distribution system, the switch action path represents the topological state of each stage of power supply restoration, and the target power supply restoration model is constructed based on the goal of minimizing load value loss; According to the switch action path and the target power supply restoration model, a power supply restoration method is solved for the sub-area to obtain a target restoration method.

2. The method according to claim 1, characterized in that: The determining of the switch action path based on the pre-built Monte Carlo tree and the target power supply restoration model includes: In each round of iteration, path exploration is performed based on the Monte Carlo tree to obtain leaf nodes to be processed; Performing simulation operation processing on the leaf node to be processed, and determining the value of the leaf node to be processed according to the processing result and the target power supply recovery model; The Monte Carlo tree is obtained according to the value update of the leaf node to be processed, and the iteration is terminated when the number of iterations meets a preset condition, and the search path of the Monte Carlo tree is determined as the switch action path.

3. The method according to claim 2, characterized in that The path exploration process is performed based on the Monte Carlo tree to obtain a leaf node to be processed, including: For any node in the Monte Carlo tree, when the node has not been subjected to path exploration processing, an expansion operation is performed on the node to obtain a plurality of child nodes; Calculating an upper confidence limit value of each of the subnodes; the upper confidence limit value is used to evaluate the estimated value of each of the subnodes; The child node with the maximum upper confidence limit value is determined as the leaf node to be processed.

4. The method according to claim 3, characterized in that The node is extended to obtain multiple child nodes, including: Enumerating the topological states of the nodes to obtain a plurality of initial switch action plans; Determining multiple feasible switching action plans according to each of the initial switching action plans and predetermined state constraints; Each of the feasible switch action schemes is executed on the node respectively, and the topological state after each of the feasible switch action schemes is executed is determined as one of the sub-nodes.

5. The method according to claim 3, characterized in that: The performing simulation operation processing on the leaf node to be processed and determining the value of the leaf node to be processed according to the processing result and the target power supply recovery model includes: Obtaining initial parameters of the leaf node to be processed; Inputting the initial parameters into the target power supply restoration model to solve the power supply problem, and obtaining the load value restoration amount of the leaf node to be processed; The load value recovery amount of the leaf node to be processed is input into a pre-built value calculation model to perform load value calculation processing to obtain the value of the leaf node to be processed.

6. The method according to claim 5, characterized in that The process of constructing the target power supply restoration model includes: Based on the pre-acquired distribution system parameter information, an initial power supply restoration model is constructed; The target power supply restoration model is constructed according to the initial power supply restoration model and the solution model.

7. The method according to claim 6, characterized in that The power distribution system parameter information includes at least one of the multiple nodes in the sub-area, the total active load, the total reactive load, and the load importance weight value of each node; the initial power supply restoration model is constructed based on the pre-acquired power distribution system parameter information, including: According to the power distribution system parameter information, determine the objective function and the first constraint condition; the first constraint condition includes a safe operation constraint, a power flow constraint and a distributed power supply operation constraint; the objective function is based on the total active load, the total reactive load and the load importance weight value; the objective function is used to characterize the priority of different loads in power supply restoration; The initial power supply restoration model is composed of the objective function and the first constraint condition.

8. The method according to claim 7, characterized in that The step of constructing the target power supply restoration model according to the initial power supply restoration model and the solution model includes: Based on the node topology state of the Monte Carlo tree, performing variable assignment processing on the initial power supply restoration model to obtain an assigned initial power supply restoration model; The target power supply restoration model is composed of the assigned initial power supply restoration model and the solution model.

9. The method according to claim 8, characterized in that The process of constructing the solution model includes: Decomposing the objective function to obtain a plurality of sub-functions; the sub-functions are composed of iteration variables and dual variables; Constructing a convergence condition based on the pre-acquired convergence parameters; the convergence parameters include the current boundary variables and the previous K boundary variables of the iteration variables; The solution model is composed of the pre-acquired second constraint condition, each of the sub-functions and the convergence condition.

10. A device for determining a distributed power supply restoration mode of a distribution network, characterized in that: The distribution network distributed power supply restoration mode determination device comprises: A region acquisition module, used for acquiring a plurality of pre-divided sub-regions of an active power distribution system; A path determination module is used to determine the switch action path for each of the sub-areas based on a pre-constructed Monte Carlo tree and a target power supply restoration model; the nodes of the Monte Carlo tree represent the topological state of the active power distribution system, the switch action path represents the topological state of each stage of power supply restoration, and the target power supply restoration model is constructed based on the goal of minimizing load value loss; A solution module is used to solve the power supply recovery mode of the sub-area according to the switch action path and the target power supply recovery model to obtain a target recovery mode.

Images