A load self-recovery optimization method suitable for distribution network black start

Abstract

The present invention relates to a load self-recovery optimization method suitable for black start of distribution network, comprising the following steps: S1, obtaining the available output of distributed power source according to the topological structure of distribution network at the time of power outage, and obtaining the importance of line; S2, evaluating the black start capability of distributed power source, and selecting black start power source; S3, constructing distribution network system partition optimization model, using genetic algorithm to randomly generate chromosomes to represent the partition to which each node belongs; S4, constructing load recovery optimization model within partition, based on randomly generated distribution network partition scheme, using genetic algorithm to randomly generate chromosomes to represent whether the load of each node in the partition is restored, and using the shortest path algorithm to search for the recovery path for the load to be restored, and obtaining the optimal load recovery scheme under the current partition; S5, selecting the global optimal partition scheme and load recovery scheme, and realizing parallel recovery of loads in each partition. Compared with the prior art, the present invention has the advantages of good starting effect, high efficiency, strong safety, etc.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network fault recovery, and in particular to a load self-recovery optimization method suitable for black start of a distribution network. Background Art

[0002] With the rapid growth of electricity demand, the operating conditions of the power system have become increasingly complex. In recent years, major power outages caused by extreme events have occurred frequently, seriously threatening the national energy security level. At the same time, the access of distributed resources such as wind power, photovoltaics, electric vehicles and energy storage to the distribution network has gradually made the distribution network active and flexible, providing flexible resource support for the partitioning of the distribution network after a major power outage to achieve local self-healing. Using distributed resources to support the parallel self-healing technology of distribution network partitions after faults can effectively reduce the social and economic losses caused by power outages, thereby further improving the resilience of the distribution network.

[0003] In recent years, domestic and foreign scholars have conducted in-depth research on black start schemes after power system blackouts. In response to large-scale power outages in distribution networks caused by transmission network failures, many scholars have mainly conducted research from three aspects: distribution network partition optimization method, distribution network network reconstruction optimization method, and load recovery optimization method. Regarding the distribution network partition optimization method, some scholars use complex network theory to partition the distribution network, including spectral clustering algorithm, GN splitting algorithm, and improved label propagation algorithm. However, the above methods often fail to consider the electrical characteristics of the power grid, resulting in insufficient results. Some scholars also partition the distribution network by constructing a mixed integer programming model. However, because they only consider how to partition, they do not take into account the impact of load recovery on the partition effect, resulting in suboptimal results. Since the distribution network partition is the basis of the black start process and interacts with the load recovery process, the influence of partition division and load recovery should be considered when formulating the load self-recovery plan for distribution network partition. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a load self-recovery optimization method suitable for distribution network black start.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A load self-recovery optimization method suitable for distribution network black start, comprising the following steps:

[0007] S1. Obtain the available output of distributed power sources according to the topological structure of the distribution network at the time of power outage, and obtain the line importance according to the node access load, node load importance and line restoration cost;

[0008] S2. Use the grey multi-attribute decision-making method to evaluate the black start capability of distributed power sources, and select power sources with self-starting capability and voltage and frequency regulation capability as black start power sources;

[0009] S3. Construct a distribution network system partition optimization model. The goal of optimizing the partitions of each node in the system is set to minimize the recovery time difference of each partition of the distribution network and the overall system recovery cost. A genetic algorithm is used to randomly generate chromosomes to represent the partition to which each node belongs, ensuring that each partition contains at least one black start power supply.

[0010] S4. Construct a load recovery optimization model within a partition. The optimization goal of each load recovery plan within the partition is set to maximize the important load recovery capacity and minimize the system recovery cost within the partition. Based on the randomly generated distribution network partitioning scheme, a genetic algorithm is used to randomly generate chromosomes to represent whether the load of each node in the partition has been restored, and the shortest path algorithm is used to search for the recovery path for the load to be restored, so as to obtain the optimal load recovery plan under the current partition;

[0011] S5. By considering the unified optimization of distribution network partitioning and load recovery within the partition, the globally optimal partitioning scheme and load recovery scheme are selected to achieve parallel recovery of loads in each partition.

[0012] In step S1, the calculation formula of the line restoration cost is:

[0013] Q l =α l (c l +s l )

[0014] Among them, Q l is the restoration cost of line l, c l is the charging reactive value of line l, α l Indicates whether line l is restored. When the value is 0, it means that line l has not been restored. When the value is 1, it means that line l has been restored. l is the number of switch operations of line 1.

[0015] In step S1, the calculation formula of line importance is:

[0016]

[0017] Among them, W l is the line importance of line l, and are the load sizes at nodes i and j at both ends of line l, respectively, and i, j∈[1, 2, …, N], N is the total number of nodes in the distribution network, λ is the load importance coefficient, and the load importance coefficients λ of the primary, secondary and tertiary loads are 1, 10 and 100 respectively.

[0018] In the step S2, the evaluation index of the black start capability of the distributed power source is specifically the black start characteristics of the distributed power source, including available output, start-up time, frequency and voltage regulation capability and load increase speed.

[0019] In step S2, the black start capability evaluation of distributed power sources using the grey multi-attribute decision-making method includes the following specific steps:

[0020] S21. Determine the effect measure θ according to the attribute type xy The value of is:

[0021] Benefit attributes: Cost attributes: Moderate properties:

[0022]

[0023] Among them, μ xy For plan Y y In property X x The evaluation results under x For attribute X x Specific evaluation outcome expectations under

[0024] S22, according to the effect measure θ xy Construct the decision matrix H = [θ xy ] n×m , where the decision matrix H is a decision matrix for a decision combination containing n attributes and m alternatives:

[0025] S23. Obtaining decision scores Among them, ω Hx is the weight of each attribute, and all ω Hx Add up to 1, G y In plan Y y Decision rating.

[0026] In step S3, the objective function of the distribution network system partition optimization model is expressed as:

[0027]

[0028] in, ψD is the total number of partitions, ψ dl is the total number of lines in partition d, t d is the recovery time of partition d, is the average recovery time of all partitions, ψdl is the total number of restored lines in partition d, Represents the overall system recovery cost.

[0029] In step S3, the genetic algorithm is used to randomly generate chromosomes to represent the partitions to which each node belongs:

[0030] The chromosomes in the genetic algorithm are constructed by integer coding. The total number of partitions, i.e., the coding range, is determined according to the number of black start power supplies in the distribution network system. All nodes in the system are encoded. The number of codes for each node represents the number of partitions of the node. The length of the chromosome represents the total number of nodes in the system. The chromosome is expressed as:

[0031] Chrom=[x1,x2,…,x N ]

[0032] Among them, x N Indicates the partition to which the Nth node belongs.

[0033] In step S4, the objective function of the load recovery optimization model within the partition is expressed as:

[0034]

[0035] Among them, ψ load is the total number of all loads in the partition system, and the coefficient α Li Indicates whether the load at node i is restored, λ i is the load importance coefficient of node i, P Li is the load size at node i, The cost of system recovery within the partition.

[0036] In step S4, the genetic algorithm is used to randomly generate chromosomes to represent whether the load of each node in the partition has recovered. Specifically:

[0037] The chromosome in the genetic algorithm is constructed by integer coding. The chromosome length is determined according to the number of load nodes in the distribution network partition. All nodes in the partition system are encoded. The encoding range is 0 or 1, that is, the encoding number of each node indicates whether the load at the node has been restored. 0 means that the node has not been restored, and 1 means that the load at the node has been restored. The chromosome is expressed as:

[0038] Chrom=[y1,y2,…,y M ]

[0039] Among them, y M Indicates the load recovery status at node M in the partition system.

[0040] In step S4, for the load nodes to be restored in the partition obtained by the genetic algorithm, the Dijkstra algorithm is used to search for the shortest path from the black start power supply to the load nodes to be restored according to the partition, and the shortest path from each vertex to other vertices is obtained by the Dijkstra algorithm. A greater line importance is set for the restored line to give priority to it, and the shortest path to restore the load node M is expressed as:

[0041]

[0042] Among them, ψ Ml The total number of lines through which load node M is restored.

[0043] The constraints of the objective functions of the distribution network system partition optimization model and the load recovery optimization model within the partition include:

[0044] Distributed power generation output constraints:

[0045]

[0046]

[0047] Among them, P Gk is the active power output of the kth distributed generation, Q Gk is the reactive power output of the kth distributed generation, ψ k It is the collection of all distributed power sources in the distribution network;

[0048] Power balance constraints:

[0049]

[0050] Among them, α Gk Indicates whether the kth power supply is started. A value of 0 indicates that the power supply is not started, and a value of 1 indicates that the power supply is started;

[0051] Node voltage constraints:

[0052] V i min ≤V i ≤V i max i∈ψ load

[0053] Among them, V i is the voltage value of load node i in the distribution network, V i max 、V i min are the maximum and minimum values ​​of the voltage respectively;

[0054] Line flow constraints:

[0055]

[0056] Among them, P Ll is the active power transmitted in line l, ψ L represents the set of all lines l in the distribution network, is the maximum value of the transmitted active power.

[0057] Compared with the prior art, the present invention has the following advantages:

[0058] 1. Most of the existing black start schemes are designed for transmission networks. When a transmission network fails and causes a large-scale power outage in the distribution network, the recovery of the distribution network needs to wait until the transmission network is restored. In addition, the distributed resources in the distribution network are not considered for black start at the distribution network level. The present invention flexibly utilizes the supporting role of distributed resources in the distribution network, divides the distribution network into zones, and each zone restores power to the surrounding loads through distributed power sources. The entire distribution network is restored in parallel and local self-healing is completed, which greatly reduces the fault recovery time after a large power outage, fully guarantees the continuous and reliable power supply of important loads, and reduces the power outage losses of the distribution network system.

[0059] 2. In the existing research on parallel restoration of distribution network partitions, the mutual influence between the partition scheme and the load restoration scheme is not considered, and the separation of the two will cause the final black start scheme to fall into the local optimum. The present invention fully considers the relationship between the partition division and the load self-healing scheme within the partition, optimizes the two in a unified manner, and uses genetic algorithms to solve the joint optimization model, and finally obtains the optimal solution for parallel restoration of distribution network partitions, with good recovery effect and high efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is the IEEE33-node distribution network diagram.

[0061] Figure 2 This is a diagram of the distribution network partition restoration plan.

[0062] Figure 3 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0063] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Note that the following embodiments are merely illustrative in nature, and the present invention is not intended to limit its applicable objects or uses, and the present invention is not limited to the following embodiments.

[0064] like Figure 3 As shown, the present invention provides a load self-recovery optimization method suitable for distribution network black start, the method comprising the following steps:

[0065] S1: Obtain the available output of distributed generation based on the topological structure of the distribution network at the time of power outage, and obtain the importance of the line based on the node access load, node load importance and line restoration cost;

[0066] S2: Use the grey multi-attribute decision-making method to evaluate the black start capability of distributed power sources, and select power sources with self-starting capability and voltage and frequency regulation capability as black start power sources;

[0067] S3: By establishing a distribution network system partition optimization model, the goal of optimizing the partitions of each node in the system is set to minimize the recovery time difference of each partition of the distribution network and the overall system recovery cost. A genetic algorithm is used to randomly generate chromosomes to represent the partition to which each node belongs, ensuring that each partition contains at least one black start power supply;

[0068] S4: By establishing a load recovery optimization model within a partition, the optimization goal of each load recovery plan within the partition is set to maximize the important load recovery capacity and minimize the system recovery cost within the partition. Based on the randomly generated distribution network partitioning scheme, a genetic algorithm is used to randomly generate chromosomes to represent whether the load of each node in the partition has been restored, and the shortest path algorithm is used to search for the recovery path for the load to be restored, and the optimal load recovery plan under the current partition is obtained;

[0069] S5: By considering the unified optimization of distribution network partitioning and load recovery within the partition, the globally optimal partitioning scheme and load recovery scheme are selected to achieve parallel recovery of loads in each partition.

[0070] In step S1, the calculation formula of the line restoration cost is:

[0071] Q l =α l (c l +s l )

[0072] Among them, Q l is the restoration cost of line l, c l is the charging reactive value of line l, α l Indicates whether line l is restored. When the value is 0, it means that line l has not been restored. When the value is 1, it means that line l has been restored. l is the number of switch operations of line 1.

[0073] In step S1, the calculation formula of line importance is:

[0074]

[0075] Among them, the line importance of line l is given by W l express, and They represent the load sizes at nodes i and j at both ends of line l, respectively, where i and j∈[1, 2, …, N], N is the total number of nodes in the distribution network, and λ represents the load importance coefficient, where the load importance coefficients λ of the first, second and third level loads are 1, 10 and 100 respectively.

[0076] In step S2, the capability evaluation index of the distributed power supply black start is specifically:

[0077] The black start characteristics of distributed generation are selected, including available output, start-up time, frequency and voltage regulation capability, and load ramp-up speed, to construct an attribute set for black start capability evaluation decision.

[0078] In step S2, the evaluation process of grey multi-attribute decision-making is divided into three steps, including effect measurement, building a decision matrix and decision scoring.

[0079] Step S21, effect measurement θ xy The value of needs to be evaluated according to the attribute type:

[0080] Benefit attributes Cost attribute Moderate attributes

[0081] Among them, μ xy Represents any solution Y y In any attribute X x The evaluation results under θ x ′ is attribute X x Specific evaluation outcome expectations.

[0082] Step S22: construct a decision matrix H = [θ xy ] n×m , where H is a decision matrix established for a decision combination containing n attributes and m alternatives.

[0083] Step S23: Decision scoring Among them, ω Hx is the weight of each attribute, all ω Hx Add up to 1, G y In plan Y y Decision rating.

[0084] In step S3, the objective function of the distribution network system partition optimization model is:

[0085]

[0086] in, ψD Indicates the total number of partitions, t d Indicates the recovery time of partition d, represents the average recovery time of all partitions,ψdl represents the total number of restored lines in partition d, Represents the overall system recovery cost.

[0087] In step S3, the genetic algorithm specifically includes:

[0088] The chromosomes in the genetic algorithm are constructed using integer coding. The total number of partitions, i.e., the coding range, is determined according to the number of black start power supplies in the distribution network system. All nodes in the system are encoded. The number of codes for each node represents the number of partitions of the node, and the length of the chromosome represents the total number of nodes in the system. The chromosome can be expressed as:

[0089] Chrom=[x1,x2,…,x N ]

[0090] Among them, x N Indicates the partition to which the Nth node belongs. Assuming that the 6-node distribution network contains 2 black start power supplies, the chromosome can be expressed as Chrom = [1, 2, 2, 1, 2, 1], indicating that nodes 1, 4, 6 belong to partition 1, and the remaining nodes belong to partition 2.

[0091] In step S4, the objective function of the load recovery optimization model within the partition is:

[0092]

[0093] Among them, ψ load represents the total number of all loads in the partition system, and the coefficient α Li Indicates whether the load at node i is restored, λ i represents the load importance coefficient of node i, P Li represents the load size at node i, The cost of system recovery within the partition.

[0094] In step S4, the genetic algorithm specifically includes:

[0095] The chromosome in the genetic algorithm is constructed using integer coding. The chromosome length is determined according to the number of load nodes in the distribution network partition. All nodes in the partition system are encoded, and the encoding range is 0 or 1. That is, the encoding number of each node indicates whether the load at the node has been restored. 0 means that the node has not been restored, and 1 means that the load at the node has been restored. The chromosome can be expressed as:

[0096] Chrom=[y1,y2,…,y M ]

[0097] Among them, y MIndicates the load recovery status at node M in the partition system. Assuming a 6-node partition system, the chromosome can be expressed as Chrom = [1, 0, 0, 1, 0, 1], indicating that the loads at nodes 1, 4, and 6 have been restored, while the loads at the remaining nodes have not been restored.

[0098] In step S4, the shortest path algorithm, Dijkstra algorithm, is used to search for the shortest path from the black start power supply to the load nodes to be restored in the partitions obtained by the genetic algorithm. The idea of ​​Dijkstra algorithm is to find the shortest path from each vertex to other vertices. Let L(v i ,v k ) represents point v i To point v k The shortest path, L(v k ,v j ) represents point v k To point v j The shortest path, L(v i ,v j ) represents point v i To point v j The path of point v i To point v j The shortest path is:

[0099] min(L(v i ,v k )+L(v k ,v j ),L(v i ,v j )),v k∈ ψ load

[0100] For restored lines, a greater line importance is set to give priority to the restored lines. The shortest path to restore the load node M can be expressed as:

[0101]

[0102] Among them, ψ Ml The total number of lines through which load node M is restored.

[0103] In step S3 and step S4, the constraints of the objective functions of the distribution network system partition optimization model and the load recovery optimization model within the partition include distributed power generation output constraints, power balance constraints, node voltage constraints, and line flow constraints, specifically:

[0104] Distributed power generation output constraints:

[0105]

[0106]

[0107] Among them, P Gk is the active power output of the kth distributed generation, Q Gk is the reactive power output of the kth distributed generation, ψ k It is the collection of all distributed power sources in the distribution network;

[0108] Power balance constraints:

[0109]

[0110] Among them, α Gk Indicates whether the kth power supply is started. A value of 0 indicates that the power supply is not started, and a value of 1 indicates that the power supply is started;

[0111] Node voltage constraints:

[0112] V i min ≤V i ≤V i max i∈ψ load

[0113] Among them, V i is the voltage value of load node i in the distribution network, V i max 、V i min are the maximum and minimum values ​​of the voltage respectively;

[0114] Line flow constraints:

[0115]

[0116] Among them, P Ll is the active power transmitted in line l, ψ L represents the set of all lines l in the distribution network, is the maximum value of the transmitted active power.

[0117] Example

[0118] In this embodiment, Figure 1 The IEEE33 distribution network system shown in the figure is used as an example to perform load partition self-recovery based on the method of the present invention. The distribution network system includes 7 distributed power sources, which are connected to nodes 1, 7, 13, 17, 21, 25, and 30 respectively. The specific data of the distributed power sources are shown in Table 1.

[0119] Table 1 Distributed power supply parameters

[0120]

[0121] In the following text, DG is used to represent distributed power sources. From the black start capability decision scoring results, it can be seen that DG5, DG6 and DG7 have relatively higher scores and are suitable for selection as black start power sources to start first. Therefore, it is preliminarily determined that the distribution network is divided into 3 partitions for parallel recovery. It is known that the loads at nodes 3, 4, 8, 10, 18, 20, 24, 25, 31, and 33 of the system are primary loads. Therefore, when formulating the partition division plan and load recovery plan, the restoration of power supply to important loads should be considered first. The loads at nodes 5, 9, 12, 15, 17, 21, 23, 28, 29, and 32 of the system are secondary loads. Therefore, after the primary loads resume normal power supply, priority is given to restoring the secondary load nodes, and finally restoring other load nodes according to the remaining available output of each partition power supply.

[0122] For the load self-recovery strategy scheme of distribution network partition, according to Figure 2 The results were analyzed:

[0123] Firstly, based on the distribution network system partition optimization model, a genetic algorithm is used to partition all load nodes, and finally divided into three partitions. Then, based on the load recovery optimization model within the partition, a genetic algorithm is used to solve the load nodes to be restored within the partition. Finally, the shortest path algorithm is used to search for the recovery path for the load to be restored, and the optimal load recovery plan under the current partition is obtained.

[0124] The power sources included in zone 1 are DG1 and DG5, with a total available output of 1000kW, including load nodes 1, 2, 3, 4, 5, 19, 23, and 24, with a total power loss load of 970kW. The primary loads are nodes 3, 4, and 24, and the secondary loads are nodes 5 and 23. The total power output can meet the needs of restoring power to all loads. Based on the restoration path determined by the shortest path algorithm, the power supply of the primary load nodes 3, 4, and 24 is restored first, followed by the power supply of the secondary load nodes 5 and 23, and finally the other load nodes.

[0125] The power sources included in zone 2 are DG2, DG3 and DG7, with a total available output of 1470kW, including load nodes 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 20, 21, 22, and a total power loss load of 1495kW. The first-level loads are nodes 8, 10, 18, 20, and the second-level loads are nodes 9, 12, 15, 17, 21. The total power output is not enough to restore the power supply of all loads. Based on the restoration path determined by the shortest path algorithm, the power supply of the first-level load nodes 8, 10, 18, 20 is restored first, followed by the power supply of the second-level load nodes 9, 12, 15, 17, 21, and finally the other load nodes. Since the total power output is not enough to restore the power supply of all loads, the load node 11 with a lower importance level and smaller capacity is not restored.

[0126] The power sources included in zone 3 are DG4 and DG6, with a total available output of 1300kW, including load nodes 25, 26, 27, 28, 29, 30, 31, 32, and 33, with a total power loss load of 1340kW. The first-level loads are nodes 25, 31, and 33, and the second-level loads are nodes 28, 29, and 32. The total power output of the power source is not enough to restore the power supply of all loads. Based on the restoration path determined by the shortest path algorithm, the power supply of the first-level load nodes 25, 31, and 33 is restored first, followed by the power supply of the second-level load nodes 28, 29, and 32, and finally the other load nodes. Since the total power output is not enough to restore the power supply of all loads, the load node 26 with a lower importance level, smaller capacity, and a larger restoration path cost is not restored.

[0127] Obviously, by adopting the method of the present invention, the loss caused by a major power outage in the distribution network can be greatly reduced, the overall recovery process of the black start can be accelerated, and the resilience level of the distribution network can be improved.

[0128] The above embodiments are merely examples and do not limit the scope of the present invention. These embodiments can also be implemented in various other ways, and various omissions, substitutions, and changes can be made without departing from the technical concept of the present invention.

Claims

1. A load self-recovery optimization method suitable for distribution network black start, characterized in that: The following steps are involved: S1. Obtain the available output of distributed power sources according to the topological structure of the distribution network at the time of power outage, and obtain the line importance according to the node access load, node load importance and line restoration cost; S2. Use the grey multi-attribute decision-making method to evaluate the black start capability of distributed power sources, and select power sources with self-starting capability and voltage and frequency regulation capability as black start power sources; S3. Construct a distribution network system partition optimization model. The goal of optimizing the partitions of each node in the system is set to minimize the recovery time difference of each partition of the distribution network and the overall system recovery cost. A genetic algorithm is used to randomly generate chromosomes to represent the partition to which each node belongs, ensuring that each partition contains at least one black start power supply. S4. Construct a load recovery optimization model within a partition. The optimization goal of each load recovery plan within the partition is set to maximize the important load recovery capacity and minimize the system recovery cost within the partition. Based on the randomly generated distribution network partitioning scheme, a genetic algorithm is used to randomly generate chromosomes to represent whether the load of each node in the partition has been restored, and the shortest path algorithm is used to search for the recovery path for the load to be restored, so as to obtain the optimal load recovery plan under the current partition; S5. By considering the unified optimization of distribution network partitioning and load recovery within the partition, the globally optimal partitioning scheme and load recovery scheme are selected to achieve parallel recovery of loads in each partition.

2. A load self-recovery optimization method suitable for distribution network black start according to claim 1, characterized in that: In step S1, the calculation formula of the line restoration cost is: Q l =α l (c l +s l ) Among them, Q l is the restoration cost of line l, c l is the charging reactive value of line l, α l Indicates whether line l is restored. When the value is 0, it means that line l has not been restored. When the value is 1, it means that line l has been restored. l is the number of switch operations of line l; The calculation formula of line importance is: Among them, W l is the line importance of line l, and are the load sizes at nodes i and j at both ends of line l, respectively, and i, j∈[1, 2, …, N], N is the total number of nodes in the distribution network, λ is the load importance coefficient, and the load importance coefficients λ of the primary, secondary and tertiary loads are 1, 10 and 100 respectively.

3. The load self-recovery optimization method applicable to distribution network black start according to claim 1, characterized in that: In the step S2, the evaluation index of the black start capability of the distributed power source is specifically the black start characteristics of the distributed power source, including available output, start-up time, frequency and voltage regulation capability and load increase speed.

4. The load self-recovery optimization method applicable to distribution network black start according to claim 1 is characterized in that: In step S2, the black start capability evaluation of distributed power sources using the grey multi-attribute decision-making method includes the following specific steps: S21. Determine the effect measure θ according to the attribute type xy The value of is: Benefit attributes: Cost attributes: Moderate properties: Among them, μ xy For plan Y y In property X x The evaluation results under θ x ′ is attribute X x Specific evaluation outcome expectations under S22, according to the effect measure θ xy Construct the decision matrix H = [θ xy ] n×m , where the decision matrix H is a decision matrix for a decision combination containing n attributes and m alternatives: S23. Obtaining decision scores Among them, ω Hx is the weight of each attribute, and all ω Hx Add up to 1, G y In plan Y y Decision rating.

5. The load self-recovery optimization method applicable to distribution network black start according to claim 2 is characterized in that: In step S3, the objective function of the distribution network system partition optimization model is expressed as: in, ψD is the total number of partitions, t d is the recovery time of partition d, is the average recovery time of all partitions, ψdl is the total number of restored lines in partition d, Represents the overall system recovery cost.

6. A load self-recovery optimization method suitable for distribution network black start according to claim 5, characterized in that: In step S3, the genetic algorithm is used to randomly generate chromosomes to represent the partitions to which each node belongs: The chromosomes in the genetic algorithm are constructed by integer coding. The total number of partitions, i.e., the coding range, is determined according to the number of black start power supplies in the distribution network system. All nodes in the system are encoded. The number of codes for each node represents the number of partitions of the node. The length of the chromosome represents the total number of nodes in the system. The chromosome is expressed as: Chrom=[x1,x2,…,x N ] Among them, x N Indicates the partition to which the Nth node belongs.

7. The load self-recovery optimization method applicable to distribution network black start according to claim 3 is characterized in that: In step S4, the objective function of the load recovery optimization model within the partition is expressed as: Among them, ψ load is the total number of all loads in the partition system, and the coefficient α Li Indicates whether the load at node i is restored, λ i is the load importance coefficient of node i, P Li is the load size at node i, The cost of system recovery within the partition.

8. A load self-recovery optimization method suitable for distribution network black start according to claim 7, characterized in that: In step S4, the genetic algorithm is used to randomly generate chromosomes to represent whether the load of each node in the partition has recovered. Specifically: The chromosome in the genetic algorithm is constructed by integer coding. The chromosome length is determined according to the number of load nodes in the distribution network partition. All nodes in the partition system are encoded. The encoding range is 0 or 1, that is, the encoding number of each node indicates whether the load at the node has been restored. 0 means that the node has not been restored, and 1 means that the load at the node has been restored. The chromosome is expressed as: Chromium=[y1,y2,…,y M ] Among them, y M Indicates the load recovery status at node M in the partition system.

9. A load self-recovery optimization method suitable for distribution network black start according to claim 8, characterized in that: In step S4, for the load nodes to be restored in the partition obtained by the genetic algorithm, the Dijkstra algorithm is used to search for the shortest path from the black start power supply to the load nodes to be restored according to the partition, and the shortest path from each vertex to other vertices is obtained by the Dijkstra algorithm. A greater line importance is set for the restored line to give priority to it, and the shortest path to restore the load node M is expressed as: Among them, ψ Ml The total number of lines through which load node M is restored.

10. A load self-recovery optimization method suitable for distribution network black start according to claim 5 or 7, characterized in that: The constraints of the objective functions of the distribution network system partition optimization model and the load recovery optimization model within the partition include: Distributed power generation output constraints: Among them, P Gk is the active power output of the kth distributed generation, Q Gk is the reactive power output of the kth distributed generation, ψ k It is the collection of all distributed power sources in the distribution network; Power balance constraints: Among them, α Gk Indicates whether the kth power supply is started. A value of 0 indicates that the power supply is not started, and a value of 1 indicates that the power supply is started; Node voltage constraints: In i min ≤V i ≤V i max i∈ψ load Among them, V i is the voltage value of load node i in the distribution network, V i max 、V i min are the maximum and minimum values ​​of the voltage respectively; Line flow constraints: Among them, P Ll is the active power transmitted in line l, ψ L represents the set of all lines l in the distribution network, is the maximum value of the transmitted active power.

Images