An efficient self-healing method for distribution network after fault based on feeder grouping and clustering

Abstract

The present invention discloses a method for efficient self-healing after a distribution network fault based on feeder grouping and clustering, which relates to the field of distribution network optimization and safety, and includes the following steps: Step 1, traversing the power supply path of the distribution network; Step 2, constructing a load recovery model for the distribution network after a fault based on the power supply path, minimizing the overload rate and load loss rate of the substation; Step 3, using the load chain and load block theory to quantify the importance of each interconnecting switch in the distribution network; Step 4, using the optimal clustering algorithm based on mixed integer linear programming, defaulting redundant switches with lower importance; Step 5, launching distributed fault self-healing for each sub-feeder group, and verifying the effectiveness of feeder grouping and clustering. The innovative introduction of load chain theory and distribution network power supply path search method realizes the quantification of the importance of complex distribution network interconnecting switches in a simple manner without the need for flow calculation; the default of redundant interconnecting switches and feeder grouping are realized, and the calculation speed of switch action plans is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of distribution network optimization and safety, and in particular to a high-efficiency self-healing method for distribution network after fault based on feeder grouping and clustering. Background Art

[0002] In modern society, electricity has become an indispensable key energy source to maintain the normal operation of cities. With the continuous advancement of global urbanization, urban population continues to grow, and the scale of industrial production continues to expand, accompanied by a rapid increase in urban electricity demand. In this context, the distribution network, as a key link between the power system and end users, has attracted much attention for its power supply reliability. However, due to the complex grid structure, numerous equipment, and harsh operating environment, the distribution network has a high frequency of failures, which brings many inconveniences to production and life. Therefore, it is necessary to build a fast load transfer system after the distribution network fails to achieve efficient self-healing of the fault. The existing distribution network fault self-healing process mainly relies on the coordination of the contact switch and the section switch to transfer the load of the fault section to the remaining substations as much as possible to achieve the maximum load restoration.

[0003] Considering the large scale and complex system of the distribution network, solving the global fault self-healing problem has shortcomings such as slow speed and low efficiency. Therefore, it is necessary to propose corresponding methods to simplify the complexity and accelerate the calculation of the distribution network fault self-healing problem. Most of the existing studies rely on the distribution network line impedance parameters that can be accurately obtained. This requirement is contrary to the "extensive" characteristics of the actual distribution network, so there are many limitations in engineering applications. In addition, it is too resource-consuming and unnecessary to achieve optimal power restoration through the coordination of all switches in the entire distribution network. In summary, the existing research lacks an effective and practical strategy for simplifying the complexity and accelerating the calculation of the distribution network fault self-healing problem. Summary of the invention

[0004] The object of the present invention is to overcome one or more deficiencies of the prior art and to provide a method for self-healing a distribution network after a fault based on feeder grouping and clustering.

[0005] The objective of the present invention is achieved through the following technical solutions:

[0006] A method for efficient self-healing of a distribution network after a fault based on feeder grouping and clustering is provided, comprising the following steps:

[0007] Step 1: Based on the improved depth-first algorithm, a distribution network power supply path search algorithm is established to traverse all feasible power supply paths of the distribution network;

[0008] Step 2: Construct a load recovery model after a distribution network fault based on the power supply path to minimize the overload rate and load loss rate of the substation;

[0009] Step 3: By applying the load chain and load block theory and combining the power supply path, the importance of each tie switch in the distribution network is quantified;

[0010] Step 4: By using the optimal clustering algorithm based on mixed integer linear programming, the redundant switches with lower importance are defaulted to decouple the entire distribution network into multiple sub-feeder groups;

[0011] Step 5: Carry out distributed fault self-healing for each sub-feeder group and verify the effectiveness of feeder grouping and clustering;

[0012] In step 1, a distribution network power supply path search algorithm is established. First, the distribution network is physically modeled. Specifically, the distribution network is regarded as a network consisting of several nodes. With several edges Topology diagram ; The nodes of the distribution network include power nodes and load nodes; the edges are the connecting lines between two nodes, including the non-switch branch, the section switch branch and the connecting switch branch between the feeder; the distribution network graph model Using a node adjacency matrix To characterize, each element in the matrix The calculation method is as follows:

[0013] ;

[0014] Among them, when , There is a connecting branch between two nodes. When, element ; When there is no connecting branch between two nodes, the element ;

[0015] The step 5 finally carries out distributed fault self-healing for each sub-feeder group and verifies the effectiveness of feeder grouping and clustering: After steps 1 to 4, the original complex distribution network has been divided into several independent feeder groups. Therefore, when a fault occurs somewhere, the original distribution network The global load transfer problem can be simplified to the feeder group to which the fault belongs The internal load transfer problem is solved, while the operating status of the other feeder groups remains unchanged, that is, the following formula:

[0016] .

[0017] Furthermore, the distribution network power supply path search algorithm step also includes:

[0018] The power supply path of the distribution network is defined as follows: from the power node Starting from the power node, a non-connected path is reached through a series of edges to any load node; The set of all power supply paths starting from , among which, there are Power supply path, The power supply path is ;right Search and traverse; the calculation steps are as follows S21-S24:

[0019] S21: First define a stack Used to store the nodes that the power supply path passes through in order and store the power supply nodes Push into the stack and become the top node ; Adjacency matrix based on distribution network graph model , search for the current top node of the stack The set of adjacent nodes ;

[0020] S22: Then, the following measures are taken for the adjacent nodes in the set: determine whether the node satisfies the three criteria for stacking at the same time, if so, it is stacked and becomes the new top of the stack, and the power supply path of the stack from bottom to top is connected. Included in path collection If not satisfied, jump to the next adjacent node for judgment; adjacent node The criteria for stacking are:

[0021] 1) The adjacent node is not in the existing stack;

[0022] 2) The adjacent node is not a power node;

[0023] 3) Existing path collection This new path is not included;

[0024] S23: If the current top node of the stack If there is no adjacent node that satisfies all three conditions, Execute the pop operation;

[0025] S24: Yes After the stack is popped, the adjacent nodes of the new top node are determined, and the determination process is the same as above.

[0026] Furthermore, the load recovery model after the distribution network failure based on the power supply path in step 2; firstly, the relevant variables of the model are defined as follows: Load Node For all the end nodes A collection of power supply paths, For load No. power supply paths; By all the power supply paths starting from the power node Traverse to get (based on path search algorithm); definition For passing edge All A collection of power supply paths, Similarly, all power supply paths starting from the power node Traverse to get; define binary variables and For path and The on / off status:

[0027] ,

[0028] ;

[0029] Among them, the specified path The end load is ,path The end load is ;

[0030] The theory of load supply path of distribution network includes three important properties;

[0031] Property 1: When the following two constraints are met, the topology of the distribution network is radial, which proves that the existing transfer scheme is feasible:

[0032] ,

[0033] ;

[0034] Property 1 For any load node : Among all the power supply paths ending at this node, there is only one path;

[0035] Property 2: Distribution network power node The load carried is:

[0036] ;

[0037] Property 2 If the path is a path, then it is included in Any path within It is also a path; relevant research has proved that this property is a necessary and sufficient condition for the interconnection switch transfer scheme to meet the "open-loop operation-radiation interconnection" of the distribution network; Property 2 is used to calculate the load rate of the substation node in the distribution network, so as to judge the current substation Is there any overload phenomenon?

[0038] Property 3: The current carrying capacity of distribution line m is:

[0039] ;

[0040] Based on the above three properties, the optimal scheduling model for load recovery after distribution fault is obtained as follows:

[0041] ;

[0042] ;

[0043] ;

[0044] ;

[0045] ;

[0046] in, Indicates load block The cutting load, Represents a substation node The rated active capacity, Indicates line The objective function is to minimize the sum of load reduction and substation load rate. The constraints are used to ensure that the topology of the distribution network is always in an open-loop radiation state, that there is no overload in the substations in the distribution network, that there is no overcurrent in the tie lines, and that the load reduction of each node shall not be higher than its load size.

[0047] Furthermore, the step 3 of quantifying the importance of each tie switch in the distribution network includes: setting the load chain from the power supply node Departure, with contact switch branch is the terminal branch and only passes through The set of all power supply paths of a tie switch The minimum distance path in ; The load block is the load between two switch branches, which is the node load in the calculation network; the independent load block is the load block of the independent power supply part of each load chain in the distribution network, which is an independent load block. An independent load block corresponds to a load chain and a tie switch branch;

[0048] Regarding the identification method of load chains and independent load blocks; first, the power node is obtained through the improved path search algorithm in step 1 The starting power supply path collection ; Secondly, extract China-Israel The branch is a terminal branch and only passes through A subset of the power supply paths of a tie switch Finally, The shortest path in the power node —Tie switch "The corresponding load chain is as follows:

[0049] ;

[0050] Assume that the substation node Power supply shared load chains, corresponding to Tie switch , tie switch branch Corresponding slave substation node The starting load chain is ; Get from the power node Departure Load Chain Collection ; Among them, "substation node —Tie switch "Corresponding independent power supply path for:

[0051] ;

[0052] The substation node —Tie switch "Corresponding independent load block Independent path The set of contained load blocks;

[0053] Finally, based on the load chain theory, the importance of each tie switch in the distribution network is quantified: for a single power node in the distribution network For example, when a fault occurs at the T-junction, the transfer scale required to achieve fault self-healing and load recovery is the largest; considering that the independent load block The transfer of supply can only rely on a single load chain And the corresponding contact switch , then the size of the independent load block is equal to the maximum load recovery capacity of its corresponding tie switch, which is the importance of this switch in the fault self-healing process;

[0054] ;

[0055] In addition, there is often more than one power supply node (substation) in a conventional distribution network; considering the existence of multiple power supply nodes, it is necessary to analyze the independent load blocks on both sides of the tie switch to determine the impact range of the switch and construct its comprehensive load recovery capability (switch importance); assuming that there is Power Node , for the first Contact switch , its comprehensive load recovery capability and importance for:

[0056] ;

[0057] in, Represents the load chain passing through the tie switch of "kl" The maximum value of all independent load blocks on the side of the tie switch is The maximum influence range of one side of the node; similarly, Represents the tie switch that the load chain passes through "lk" The set of all independent load blocks on the same side also takes the maximum value.

[0058] Furthermore, the optimal clustering grouping strategy for distribution network feeders in step 4 should follow the following principles:

[0059] 1) To simplify feeder scheduling in actual projects, the number of feeders in each group should be less than a certain artificially set integer threshold. ; 2). For the tie switches with higher load recovery importance, they should be retained first (i.e., as the edge connecting the feeders within the group); 3). For the tie switches with lower load recovery importance, they should be deleted first (i.e., as the inter-group tie edges);

[0060] The feeders in the distribution network are regarded as generalized nodes , the tie switches between feeders are regarded as generalized edge Therefore, the distribution network is regarded as a new generalized graph model ; In the generalized graph model of the distribution network described in step 1, each edge Assigned a weight The weight is the importance of the tie switch to load recovery. ; An optimal feeder clustering method based on integer linear programming is proposed:

[0061] Assume that the distribution network contains Feeders, the number of feeders in each group cannot exceed the integer threshold , first define the following variables:

[0062] :If the generalized node (feeder group) Belongs to feeder group (class) ,but ,otherwise ;

[0063] :If the generalized edge (tie switch) If disconnected, , if closed then ;

[0064] :If the feeder group is clustered is used, it means that it contains a feeder, then ,otherwise ; Then construct the objective function of this feeder clustering model:

[0065] ;

[0066] Where is the objective function of feeder grouping clustering, which minimizes the sum of the weights of disconnected tie switches and preferentially disconnects tie switches with lower importance to load recovery. The constraints of the optimal clustering model for feeders are as follows:

[0067] ;

[0068] Among them, it means that each feeder, that is, each generalized node, can and must belong to only one feeder group; among them, is the clustering variable The dimension is the maximum number of feeder groups; generally , is the total number of distribution network feeders;

[0069] ;

[0070] Among them, represents: the number of feeders contained in each feeder group must be less than or equal to the artificial threshold , to simplify the feeder dispatching operation within the group in actual engineering;

[0071] ;

[0072] Among them, it is used to ensure that: if there is any feeder Divided into feeder groups , then the variable corresponding to this feeder group is , then this feeder group is used;

[0073] [ N / τ ] + 1 ≤ ∑ k = 1 K y k ≤ N ;

[0074] Among them, the number of feeder groups obtained by clustering is limited in and between The inside is the rounding operator;

[0075] ;

[0076] Among them, it is necessary to ensure the connectivity of the feeder within the class, that is, if the tie switch Feeders on both sides , All belong to the feeder group , then the connected variable corresponding to this switch ; otherwise 0; however, the quadratic term , so it is a nonlinear constraint condition; to further simplify this problem, we use auxiliary variables , used to relax the original intra-class feeder connectivity constraints;

[0077] ,

[0078] ,

[0079] ,

[0080] ,

[0081] ;

[0082] The model completely transforms the original unconventional feeder clustering problem with edge weight constraints into a linear integer optimal programming problem, which can be solved by calling any mature solver and ensures that the obtained feeder clustering result is globally optimal.

[0083] The beneficial effects of the present invention are:

[0084] (1) The load chain theory and distribution network power supply path search method are innovatively introduced to achieve a simple quantification of the importance of the tie switches in complex distribution networks without the need for power flow calculations;

[0085] (2) A feeder clustering method based on integer linear programming is proposed to achieve the default of redundant tie switches and feeder grouping;

[0086] (3) Based on the feeder grouping and clustering results, the original global fault self-healing problem of the distribution network is decoupled into distributed fault self-healing problems within multiple independent feeder groups, which greatly improves the calculation speed of the switch action plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 It is a schematic diagram of the method steps of the present invention;

[0088] Figure 2 The power distribution network power supply path search algorithm process of the present invention;

[0089] Figure 3 It is a schematic diagram of the load chain, load block and independent load block of the present invention;

[0090] Figure 4 This is a schematic diagram of a T-junction fault transfer according to the present invention;

[0091] Figure 5 This is a schematic diagram of the grouping and clustering of feeder lines in the distribution network of the present invention. DETAILED DESCRIPTION

[0092] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0093] See also Figure 1-Figure 2 , this embodiment provides a:

[0094] See also Figure 1 :Comprising the following steps:

[0095] Step 1: Based on the improved depth-first algorithm, a distribution network power supply path search algorithm is established to traverse all feasible power supply paths of the distribution network;

[0096] Step 2: construct a load recovery model after distribution network failure based on the power supply path, aiming to minimize the overload rate and load loss rate of the substation;

[0097] Step 3: Using the load chain and load block theory, combined with the power supply path, quantify the importance of each tie switch in the distribution network;

[0098] Step 4: Default the redundant switches with lower importance and decouple the entire distribution network into multiple sub-feeder groups;

[0099] Step 5: Perform distributed fault self-healing on each sub-feeder group and verify the effectiveness of feeder grouping and clustering.

[0100] See also Figure 2 , Figure 2 The distribution network power supply path search algorithm process of the present invention, the steps of tie switch sorting are: 1) First define a stack Used to store the nodes that the power supply path passes through in order and store the power supply nodes Push into the stack and become the top node Adjacency matrix based on distribution network graph model , search for the current top node of the stack The set of adjacent nodes .

[0101] 2) Then, the following measures are taken for the adjacent nodes in the set: determine whether the node meets the three criteria for stacking at the same time. If so, it is stacked and becomes the new top of the stack, and the power supply path of the stack from bottom to top is connected. Included in path collection If not satisfied, jump to the next adjacent node for determination. The criteria for stacking are: 1) not in the existing stack; 2) not a power node; 3) existing path set Does not include the new path. If the current top node If there is no adjacent node that satisfies all three conditions, Execute the pop operation. 4) After the stack is popped, the adjacent nodes of the new top node are determined, and the determination process is the same as above.

[0102] See also Figure 3 , Figure 3 The load chain, load block, and independent load block of the present invention are schematic diagrams. The load chain is from the power supply node Departure, with contact switch branch is the terminal branch and only passes through The set of all power supply paths of a tie switch The minimum distance path in . That is The load block is the load between two switch branches, that is, the node load in the calculation network. The independent load block is the load block of the independent power supply part of each load chain in the distribution network, that is, an independent load block. An independent load block corresponds to a load chain and a tie switch branch. Figure 3 Taking the simple distribution network shown as an example, it includes 4 substations F1-F4, normally open contact switches TS2-TS4 between the 3 substations, several normally closed substation section switches and load nodes. Among them, substation F1 contains 3 load chains, corresponding to 3 contact switches, namely F1-TS1, F1-TS2, and F1-TS3; the load block is each load node.

[0103] See also Figure 4 , Figure 4 This is a schematic diagram of the T-type intersection fault transfer of the present invention. For a single power node in the distribution network For example, when a fault occurs at the T-junction, the transfer scale required to achieve fault self-healing and load recovery is the largest. The transfer of supply can only rely on a single load chain And the corresponding contact switch , then the size of the independent load block is equal to the maximum load recovery capacity of the corresponding tie switch, that is, the importance of the switch in the fault self-healing process. Figure 4 For example, when a fault occurs at the T-junction of the distribution network, the transfer scale required to achieve load recovery is the largest. Among them: the independent load block corresponding to F1-TS3 must be transferred to F3, and the independent load block corresponding to F1-TS4 must be transferred to F4. Therefore, the size of the independent load block corresponds to the maximum load recovery capacity of the tie switch, that is, the importance of the tie switch.

[0104] See also Figure 5 , Figure 5 This is the optimal clustering grouping strategy for the distribution network feeder of the present invention. Specifically, the feeders in the distribution network are regarded as generalized nodes , the tie switches between feeders are regarded as generalized edge Therefore, the distribution network is regarded as a new generalized graph model. In this embodiment Figure 5 For example, in this generalized graph model, each edge Assigned a weight The weight is the importance of the tie switch to load recovery. ,Then, a mixed integer linear programming algorithm is used to solve the unconventional feeder clustering problem with edge weight constraints. Based on the feeder grouping clustering results, the original distribution network global fault self-healing problem is decoupled into a distributed fault self-healing problem in multiple independent feeder groups, which greatly improves the calculation speed of the switch action plan.

[0105] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art shall not deviate from the spirit and scope of the present invention, and shall be within the scope of protection of the claims attached to the present invention.

Claims

1. An efficient self-healing method for distribution network after fault based on feeder grouping and clustering, characterized in that: The following steps are involved: Step 1: Establish a distribution network power supply path search algorithm to traverse all feasible power supply paths in the distribution network; Step 2: Construct a load recovery model after a distribution network fault based on the power supply path to minimize the overload rate and load loss rate of the substation; Step 3: Quantify the importance of each tie switch in the distribution network by using load chains and load blocks, combined with the power supply path; Step 4: By using the optimal clustering algorithm based on mixed integer linear programming, the redundant switches with lower importance are defaulted to decouple the entire distribution network into multiple sub-feeder groups; Step 5: Carry out distributed fault self-healing for each sub-feeder group and verify the effectiveness of feeder grouping and clustering; In step 1, a distribution network power supply path search algorithm is established. First, the distribution network is physically modeled. Specifically, the distribution network is regarded as a network consisting of several nodes. With several edges Topology diagram ; The nodes of the distribution network include power nodes and load nodes; the edges are the connecting lines between two nodes, including the non-switch branch, the section switch branch and the connecting switch branch between the feeder; the distribution network graph model Using a node adjacency matrix To characterize, each element in the matrix The calculation method is as follows: ； Among them, when , There is a connecting branch between two nodes. When, element ; When there is no connecting branch between two nodes, the element ; Step 5 performs distributed fault self-healing on each sub-feeder group and verifies the effectiveness of feeder grouping and clustering: After steps 1 to 4, the original complex distribution network is divided into several independent feeder groups. ; When a fault occurs somewhere, the original distribution network The global load transfer problem is simplified to the feeder group to which the fault belongs The internal load transfer problem is solved, while the operating status of the other feeder groups remains unchanged, that is, the following formula: 。 2. According to claim 1, a method for efficient self-healing of a distribution network after a fault based on feeder grouping and clustering, characterized in that: The distribution network power supply path search algorithm step also includes: The power supply path of the distribution network is defined as follows: from the power node Starting from the power node, a non-connected path is reached through a series of edges to any load node; The set of all power supply paths starting from , among which, there are Power supply path, The power supply path is ;right Search and traverse; the calculation steps are as follows S21-S24: S21: First define a stack Used to store the nodes that the power supply path passes through in order and store the power supply nodes Push into the stack and become the top node ; Adjacency matrix based on distribution network graph model , search for the current top node of the stack The set of adjacent nodes ; S22: Then, the following measures are taken for the adjacent nodes in the set: determine whether the node satisfies the three criteria for stacking at the same time, if so, it is stacked and becomes the new top of the stack, and the power supply path of the stack from bottom to top is connected. Included in path collection If not satisfied, jump to the next adjacent node for judgment; adjacent node The criteria for stacking are: 1) The adjacent node is not in the existing stack; 2) The adjacent node is not a power node; 3) Existing path collection This new path is not included; S23: If the current top node of the stack If there is no adjacent node that satisfies all three conditions, Execute the pop operation; S24: Yes After the stack is popped, the adjacent nodes of the new top node are determined, and the determination process is the same as above.

3. The method for efficient self-healing of a distribution network after a fault based on feeder grouping and clustering according to claim 1 is characterized in that: The load recovery model after the distribution network failure based on the power supply path in step 2; firstly, the relevant variables of the model are defined as follows: Load Node For all the end nodes A collection of power supply paths, For load No. power supply paths; By all the power supply paths starting from the power node Traverse to get; definition For passing edge All A collection of power supply paths, Similarly, all power supply paths starting from the power node Traverse to get; define binary variables and For path and The on / off status: ， ； Among them, the specified path The end load is ,path The end load is ; The theory of load supply path of distribution network includes three important properties; Property 1: When the following two constraints are met, the topology of the distribution network is radial, which proves that the existing transfer scheme is feasible: ， ； Property 1 For any load node : Among all the power supply paths ending at this node, there is only one path; Property 2: Distribution network power node The load carried is: ； Property 2 If the path is a path, then it is included in Any path within It is also a path; Property 2 is the necessary and sufficient condition for the interconnection switch transfer scheme to meet the "open-loop operation-radiation interconnection" of the distribution network; Property 2 is used to calculate the load rate of the substation node in the distribution network and determine the current substation Is there any overload phenomenon? Property 3: The current carrying capacity of distribution line m is: ； Based on the three properties, the optimal scheduling model for load recovery after distribution fault is obtained as follows: ； ； ； ； in, Indicates load block The cutting load, Represents a substation node The rated active capacity, Indicates line The objective function is to minimize the sum of load reduction and substation load rate. The constraints are used to ensure that the topology of the distribution network is always in an open-loop radiation state, that there is no overload in the substations in the distribution network, that there is no overcurrent in the tie lines, and that the load reduction of each node shall not be higher than its load size.

4. The method for efficient self-healing of a distribution network after a fault based on feeder grouping and clustering according to claim 1 is characterized in that: The step 3 of quantifying the importance of each tie switch in the distribution network includes: setting the load chain from the power supply node Departure, with contact switch branch is the terminal branch and only passes through The set of all power supply paths of a tie switch The minimum distance path in ; The load block is the load between two switch branches, which is the node load in the calculation network; the independent load block is the load block of the independent power supply part of each load chain in the distribution network, which is an independent load block. An independent load block corresponds to a load chain and a tie switch branch; Regarding the identification method of load chains and independent load blocks; first, the power node is obtained through the improved path search algorithm in step 1 The starting power supply path collection ; Secondly, extract China-Israel The branch is a terminal branch and only passes through A subset of the power supply paths of a tie switch ; Finally, The shortest path in the power node —Tie switch "The corresponding load chain is as follows: ； Assume that the substation node Power supply shared load chains, corresponding to Tie switch , tie switch branch Corresponding slave substation node The starting load chain is ; Get from the power node Departure Load Chain Collection ; Among them, "substation node —Tie switch "Corresponding independent power supply path for: ； "Substation Node —Tie switch "Corresponding independent load block Independent path The set of contained load blocks; Quantify the importance of each tie switch in the distribution network: For a single power node in the distribution network For example, when a fault occurs at the T-junction, the transfer scale required to achieve fault self-healing and load recovery is the largest. The transfer of supply can only rely on a single load chain And the corresponding contact switch , the size of the independent load block is equal to the maximum load recovery capacity of the corresponding tie switch, so the importance of this switch in the fault self-healing process is: ； There is often more than one power supply node in a conventional distribution network. Considering the existence of multiple power supply nodes, the independent load blocks on both sides of the tie switch are analyzed to determine the influence range of the switch and construct its comprehensive load recovery capability. Assuming there is Power Node , for the first Contact switch , its comprehensive load recovery capability and importance for: ； in, Represents the load chain passing through the tie switch "kl" The maximum value of all independent load blocks on the side is the sum of the independent load blocks on the side of the tie switch. The maximum influence range of one side of the node; similarly, Represents the load chain passing through the tie switch of "lk" All independent load blocks on the side are combined and the maximum value is taken.

5. The method for efficient self-healing of a distribution network after a fault based on feeder grouping and clustering according to claim 1 is characterized in that: The optimal clustering grouping strategy for distribution network feeders in step 4 is: The feeders in the distribution network are regarded as generalized nodes , the tie switches between feeders are regarded as generalized edge ; The distribution network is considered as a new generalized graph model ; In the generalized graph model of the distribution network described in step 1, each edge Assigned a weight The weight is the importance of this tie switch to load recovery. ; An optimal feeder clustering method based on integer linear programming is proposed: Assume that the distribution network contains Feeders, the number of feeders in each group cannot exceed the integer threshold , first define the following variables: :If the generalized node Belongs to feeder group ,but ,otherwise ; :If the generalized edge If disconnected, , if closed then ; :If the feeder group is clustered is used, it means that it contains a feeder, then ,otherwise ; Then the objective function of the feeder clustering model is constructed as: ； The objective function of feeder grouping clustering is to minimize the sum of the weights of disconnected tie switches, and preferentially disconnect tie switches with lower importance to load recovery. The constraints of this feeder optimal clustering model are as follows: ； Among them, each feeder, that is, each generalized node, can and must belong to only one feeder group; among them, is the clustering variable The dimension of is the maximum number of feeder groups; , is the total number of distribution network feeders; ； The number of feeders in each feeder group must be less than or equal to the artificial threshold. ; ； If there is any feeder Divided into feeder groups , then the variable corresponding to this feeder group is , then this feeder group is used: ； Among them, the number of feeder groups obtained by clustering is limited in and between The inside is the rounding operator; ； Among them, it is necessary to ensure the connectivity of the feeder within the class, that is, if the tie switch Feeders on both sides , All belong to the feeder group , then the connected variable corresponding to this switch ; otherwise 0; where the quadratic term , which is a nonlinear constraint; to further simplify this problem, we use auxiliary variables Relax the original intra-class feeder connectivity constraints; ， ， ， ， 。

Images