A multi-black-start power supply site planning method considering partition recovery

Abstract

The present application relates to power system operation control technology, in particular to a kind of multi-black start power distribution planning method considering partition recovery, this method will partition recovery model be introduced in black start power distribution optimization, consider the configuration of multiple black start power, considering black start unit position, black start unit capacity, unit start time, unit start power, path charging time and other aspects constraints, constructs the optimization model of multi-black start power distribution with the shortest black start recovery time as target;Then based on intelligent optimization algorithm, the optimal black start power distribution planning scheme is obtained by solving the optimization model of multi-black start power distribution.The method gives full play to the advantage of multi-black start power parallel recovery, improves the recovery efficiency of power system, effectively shortens the recovery time.

Description

Technical Field

[0001] This invention belongs to the field of power system operation and control technology, and specifically relates to a method for planning the deployment of multiple black-start power sources considering zone recovery. Background Technology

[0002] In recent years, large-scale power outages have continued to occur frequently both domestically and internationally due to human error, power equipment failures, and extreme weather events. Black-start power sources are fundamental to power system outage recovery, possessing self-starting capabilities in the event of a complete system blackout and providing power support for grid restoration. Hydropower and oil-fired generator units, as typical traditional black-start power sources, were the primary reliance for early power system restoration. However, these power sources are limited in capacity and their planning and construction are heavily influenced by regional conditions and natural resources, thus requiring little research on optimizing black-start power source deployment. In many regions, the reality is that the number of black-start power sources such as hydropower units is limited, while conventional thermal power generators, due to time constraints during hot starts, slow ramp-up rates, and significant impacts from starting auxiliary equipment, always require a considerable amount of time to start and output power. Optimizing the restoration control decision-making process makes it difficult to significantly improve the grid restoration speed under large-scale power outages. Black-start power sources have become a key factor restricting grid restoration efficiency; their quantity, layout, and configuration directly affect the power system restoration effect.

[0003] With the development of new technologies in the power system, various new forms of black-start power sources are gradually emerging, such as gas turbine unit retrofitting, rapid load shedding technology for thermal power units, new energy sources, and high-voltage direct current transmission. These units are less constrained by natural environmental conditions, offer greater freedom in location selection, and have larger capacities. They can serve as additional black-start power sources to provide initial generating power for system recovery, broadening the range of black-start power source capacity and spatial options. This allows for more flexible and free configuration of black-start power source capacity and location, thus meeting the power system's recovery needs after large-scale power outages and supporting rapid and reliable grid recovery. For large-scale power grids, multiple black-start power sources are often required. Planning should consider the coordination of multiple black-start power sources during recovery, taking into account the parallel recovery process across different power system zones. A scientifically sound and optimized deployment scheme will greatly accelerate the grid recovery process, shorten outage duration, improve grid recovery reliability, and enhance the power system's ability to withstand disturbances and accidents. Summary of the Invention

[0004] To address the problems existing in the background technology, this invention provides a multi-black boot power supply deployment planning method that considers partition recovery.

[0005] To solve the above technical problems, the present invention adopts the following technical solution: a method for planning the deployment of multi-black boot power supplies considering partition recovery, comprising:

[0006] Abstract the power grid topology of the black-start power source to be planned and constructed, and construct a weighted topology map of the power grid based on branch recovery time and active power; establish a power system partitioning model based on an improved label propagation algorithm;

[0007] Based on the power system zoning model, the recovery path of the units to be started in each zone is searched using the Dijkstra algorithm. Combining the unit capacity, ramp rate, starting power characteristics and the distance of the recovery path, a comprehensive index of unit start-up sequence is constructed, and a zoning recovery model of the units is established.

[0008] Based on the unit zonal recovery model, a multi-black start power supply deployment planning model for zonal recovery is established, which includes two parts: objective function and constraints. The objective function is to minimize the unit black start recovery time, and the constraints include the location of the black start unit, the capacity of the black start unit, the unit start time, the unit start power, and the path charging time.

[0009] An intelligent optimization algorithm is used to solve the multi-black boot power supply layout planning model, resulting in a multi-black boot power supply layout planning scheme that considers partition recovery.

[0010] In the above-mentioned multi-black start power supply deployment planning method considering partition recovery, the construction of the power grid weighted topology graph includes: extracting generators and loads in the power grid as indistinguishable nodes, treating lines and transformers as branches, and abstracting the power grid as a topology graph G = (V, E), where V represents the vertex set of the graph and E represents the branch set of the graph;

[0011] The recovery time of each branch and the active power flowing through each branch before the power outage are set as branch weights to obtain a weighted topology, specifically:

[0012]

[0013] In the formula, W ij |P represents the branch weight between node i and node j. ij | represents the absolute value of the active power flowing through the branch between node i and node j before the power outage, t ij This represents the recovery time of the branch between node i and node j; * indicates that the value has been normalized.

[0014] If nodes i and j have directly connected branches, according to P ij t ij The size of the value determines its weight; otherwise, the weight is 0.

[0015] In the above-mentioned multi-black-start power supply deployment planning method considering partition recovery, the establishment of a power system partition model based on the improved label propagation algorithm includes:

[0016] According to the basic requirements for power system restoration zoning, each zone must be equipped with at least one black-start power supply, and the number of zones must be consistent with the number of black-start power supplies; assuming there are n nodes and k black-start power supplies in the power grid, located at x1, x2, ..., xk nodes respectively. k The node where the black start power source is located is taken as the initial labeled node. Its label information remains unchanged throughout the propagation process. It is used to assign values ​​to the initial stage label category matrix F, specifically:

[0017]

[0018] The tag information of other non-black-start power sources and load nodes in the power grid needs to be updated and obtained during the propagation process. Therefore, except for the elements involved in the above formula, all other elements are set to zero.

[0019] The traditional label propagation algorithm is improved by introducing a partition size matrix into the propagation strategy for dividing the power system recovery partitions;

[0020] Define the partition size factor as follows:

[0021]

[0022] In the formula, m c The partition size factor for label category c, d c This represents the sum of the degrees of all nodes with label category c, where l is the total number of branches in the power grid;

[0023] The diagonal elements of the partition size matrix M are the corresponding partition size factor values, specifically:

[0024]

[0025] The M matrix represents the size of existing partitions in the system. The larger the partition size of the label, the smaller the corresponding size influence factor.

[0026] Before each round of label propagation, the current power grid partition size matrix is ​​calculated and incorporated into the considerations for node label updates;

[0027] The improved tag propagation method is as follows:

[0028]

[0029] In the formula, f(i) represents the label information of node i, C represents the type of label information, N(i) represents the set of adjacent nodes of node i, and w i (f j) indicates that the target node i and all label information are f j The sum of the edge weights of the adjacent nodes, m(f) j ) indicates that the tag information is f j Community size influencing factors;

[0030] Each node propagates according to the propagation probability and social scale matrix, specifically as follows:

[0031] Y t =F t-1 M

[0032] In the formula, the Y matrix is ​​the label influence matrix, which represents the impact of this round of propagation on the label updates of each node;

[0033] Update the label classification matrix F for this round based on the label influence matrix Y. t If the element in the c-th column of the i-th row has the largest value, then the corresponding label classification matrix F... ic The element is set to 1, and the other elements in the i-th row of the F matrix are set to 0, thereby updating the label information of each node;

[0034] The propagation process is repeated until the label classification matrix F converges, yielding the final label information and partitioning results for each node.

[0035] In the above-mentioned multi-black boot power supply deployment planning method considering partition recovery, the search for the recovery path of the unit to be booted in each partition based on the Dijkstra algorithm includes:

[0036] The connection relationships of the power grid topology are characterized by the adjacency matrix A as follows:

[0037]

[0038]

[0039] Define set V S Store the nodes where the shortest path has been found, V o Store other nodes, set D stores the distances between the black start power node and each node, matrix S stores the shortest paths between each node, and set V... S Initially, it only contained a black start power node;

[0040] According to the adjacency matrix, in V o Select the node k with the shortest distance to the black start power source and add it to set V. S Record the shortest path in matrix S, and simultaneously remove the node from set V. o Remove from;

[0041] Compare the path length obtained with node k as the intermediate node with the original path length, take the minimum of the two, and update and adjust the distance from each node in set D to the starting point.

[0042] Traverse all nodes to obtain the shortest path and shortest distance from all nodes to the black start power node, and use the shortest path as the recovery path for the node to be started.

[0043] In the above-mentioned multi-black start power supply deployment planning method considering partition recovery, the comprehensive index for constructing the unit start-up sequence includes:

[0044] Based on the unit's starting power, ramp rate, and unit capacity, the following computer group characteristic indicators are calculated:

[0045] O(k)=P * (k)-S * (k)-C * (k)

[0046] In the formula, O(k) is the characteristic index of the k-th unit, and P * (k) is the normalized value of the starting power of the k-th unit, S * (k) is the normalized value of the capacity of the k-th unit, C * (k) is the normalized value of the ramp rate of the k-th unit;

[0047] Normalization method:

[0048]

[0049] In the formula, x * Represents the normalized value of x, x min Represents the minimum value of x, x max This represents the maximum value of x;

[0050] Based on the recovery path of the unit node to be started, the sum of the branch weights contained in the recovery path is calculated and normalized as the unit distance index D(k).

[0051] Based on the unit's own characteristic indicators and distance indicators, establish comprehensive unit startup indicators:

[0052] Z(k) = O(k) + D(k)

[0053] In the formula, Z(k) is the comprehensive startup index of the k-th unit, O(k) is the characteristic index of the k-th unit, and D(k) is the distance index of the k-th unit.

[0054] In the above-mentioned multi-black boot power supply deployment planning method considering partition recovery, the objective function for establishing the multi-black boot power supply deployment planning model for partition recovery includes:

[0055] The multi-black-start power supply deployment scheme aims to minimize the black-start recovery time of units with a backup power grid. The optimization objective function can be expressed as:

[0056] minT grid =max(T) c c∈[1,2,…,k]

[0057] Among them, T grid Indicates the unit's black start recovery time; k is the number of black start power supplies, equal to the number of recovery partitions; T c Indicates the black boot time of partition C:

[0058]

[0059] In the formula, a represents the set of black-boot units within partition c, b represents the set of non-black-boot units within partition c, and t m i,BSU t represents the recovery time of the i-th black start unit. m j,NBSU This indicates the recovery time of the j-th black starter unit.

[0060] In the above-mentioned multi-black boot power supply deployment planning method considering partition recovery, the constraints of the multi-black boot power supply deployment planning model for partition recovery include:

[0061] The installation location of the black starter unit is within the node range of the power system, specifically:

[0062]

[0063] In the formula, s black,i N represents the location of the i-th black starter unit. bus This represents the set of nodes in the power grid, where n represents the number of nodes in the power grid.

[0064] The installed capacity of the black starter unit is less than the specified upper limit, specifically:

[0065] 0 <C black,i ≤C max

[0066] In the formula, C black,i C represents the installed capacity of the i-th black starter unit. max This represents the upper limit of the capacity of the black start unit;

[0067] The generator set starts within the unit's hot start time limit, and the unit resumes operation using a hot start method, specifically:

[0068]

[0069] In the formula, ts For non-black start units, t is the start-up time. s max This is the maximum hot start time limit for non-black start units, i.e., the maximum power outage time for the unit to perform a hot start.

[0070] When the unit's startup time exceeds the minimum warm-start time limit, the unit will undergo a cold start. The unit's startup time must meet the minimum cold start time limit, specifically:

[0071]

[0072] In the formula, t s The start-up time for non-black start units. Minimum cold start time limit for non-black start units;

[0073] Non-black start units rely on external power for startup. The unit starts when the system's power output meets the startup power requirement of the next unit to be started. Specifically:

[0074]

[0075] In the formula, p represents the number of black-start units in the system. Let represent the active power output by the i-th black-start generator at time t, and q represent the number of non-black-start generators that have been restored in the system. P represents the active power output by the j-th non-black start unit at time t. st This indicates the starting power required for the next unit to be started;

[0076] The optimal recovery path for unit startup is obtained using Dijkstra's algorithm, and the startup time of the unit to be started satisfies the recovery path charging time constraint:

[0077] t i,st ≥t j,st +t ij

[0078] In the formula, t i,st t represents the start-up time of unit i to be started. j,st Indicates the start-up time of unit j that has been restored, t ij This indicates the charging time of the recovery path from the restored system to the unit i to be started;

[0079] When each non-black start generator unit is started via the recovery path, the power output of the generator unit is within the upper and lower power limits, specifically:

[0080]

[0081] In the formula, P GiThis represents the active power generated by the i-th generator. This represents the minimum active power that the i-th generator is allowed to output. Q represents the maximum active power that the i-th generator is allowed to output; Gi Let be the reactive power generated by the i-th generator set. Let be the minimum reactive power that the i-th generator set is allowed to output. Let be the maximum reactive power that the i-th generator set is allowed to output;

[0082] The active power transmitted through the line should meet the upper power limit requirement, specifically:

[0083]

[0084] In the formula, P Li Let be the power flowing through the i-th line. The upper limit of power transmitted by the i-th line; the active power transmitted by the line meets the upper limit of power requirement.

[0085] Compared with existing technologies, the beneficial effects of this invention are as follows: It introduces a partitioned recovery model into the black-start power supply deployment planning, fully leveraging the advantages of parallel recovery from multiple black-start power supplies, and proposes a multi-black-start power supply deployment planning method that considers partitioned recovery. By planning and adding black-start power supplies, multi-partition parallel recovery can be performed more efficiently during the recovery process, thereby improving the power system recovery efficiency.

[0086] The multi-black-start power source deployment optimization method can distribute multiple black-start power sources more evenly on the topology network, effectively shortening the power system recovery time and controlling the duration of power outages. Attached Figure Description

[0087] Figure 1 : This is a standard computational example diagram of the New England IEEE-118 node in an embodiment of the present invention;

[0088] Figure 2 : A flowchart of a multi-black boot power supply deployment planning method considering partition recovery in an embodiment of the present invention;

[0089] Figure 3 : This is a diagram showing the calculation results of an embodiment of the present invention. Detailed Implementation

[0090] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0091] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0092] The present invention will be further described below with reference to specific embodiments, but these are not intended to limit the scope of the invention.

[0093] This embodiment proposes a multi-black-start power supply deployment planning method considering partitioned recovery. This method incorporates a partitioned recovery model into black-start power supply deployment optimization, considering the configuration of multiple black-start power supplies and taking into account constraints such as black-start unit location, capacity, startup time, startup power, and path charging time. It constructs a multi-black-start power supply deployment optimization model with the objective of minimizing black-start recovery time. An intelligent optimization algorithm is used to solve the multi-black-start power supply deployment optimization model to obtain the optimal deployment scheme. This method fully leverages the advantages of parallel recovery by multiple black-start power supplies, improves power system recovery efficiency, and effectively shortens recovery time.

[0094] This embodiment is achieved through the following technical solution: a multi-black boot power supply deployment planning method considering partition recovery, comprising the following steps:

[0095] Step 1: Abstract the power grid topology for the planned black-start power source, and construct a weighted topology map of the power grid based on branch recovery time and active power. Establish a power system partitioning model based on an improved label propagation algorithm to ensure that partition sizes are as balanced as possible, internal connections within partitions are close, and inter-partition connections are sparse.

[0096] Step 2: Based on the partitioning results obtained in Step 1, search for the recovery path of the units to be started in each partition using the Dijkstra algorithm. Combine the characteristics of unit capacity, ramp rate, startup power, and the distance of the recovery path to construct a comprehensive index of unit startup sequence and establish a unit partition recovery model.

[0097] Step 3: Based on the unit zonal recovery model obtained in Step 2, establish a multi-black start power supply layout planning model that considers zonal recovery. The model includes two parts: objective function and constraints. The objective function is to minimize the unit black start recovery time. Constraints include the location of black start units, capacity of black start units, unit start time, unit start power, and path charging time.

[0098] Step 4: Use an intelligent optimization algorithm to solve the multi-black boot power supply layout planning model described in Step 3, and obtain a multi-black boot power supply layout planning scheme that considers partition recovery.

[0099] Furthermore, the construction of the power grid topology diagram described in step 1 specifically includes:

[0100] Generators and loads in the power grid are extracted as indistinguishable nodes, and lines and transformers are regarded as branches. The power grid is abstracted as a topological graph G = (V, E), where V represents the set of vertices of the graph and E represents the set of branches of the graph.

[0101] The weights of branches are determined by comprehensively considering the recovery time of each branch and the active power flowing through each branch before the power outage, thus obtaining a weighted topology. The shorter the recovery time of a branch, the greater its weight, and the more likely the two nodes connected by the branch are to be assigned to the same partition. At the same time, the power flow data before the power outage directly reflects the tightness of the connections between various nodes in the power system. The larger the active power transmitted on a branch, the tighter the connection between the two nodes connected by the branch. Specifically:

[0102]

[0103] In the formula, W ij |P represents the branch weight between node i and node j. ij | represents the absolute value of the active power flowing through the branch between node i and node j before the power outage, t ij This represents the recovery time of the branch between node i and node j. * indicates that the value has been normalized. If nodes i and j have a directly connected branch, according to P... ij t ij The size of the value determines its weight; otherwise, the weight is 0.

[0104] Step 1, which establishes a power system partitioning model based on the improved label propagation algorithm, specifically involves:

[0105] According to the basic requirements for power system restoration zoning, each zone must be equipped with at least one black-start power source, and it is assumed that the number of zones is consistent with the number of black-start power sources. Assume there are n nodes in the power grid and k black-start power sources (located at x1, x2, ..., xk). k The node containing the black-start power source is used as the initial labeled node. Its label information remains unchanged throughout the propagation process. This is used to assign values ​​to the initial label category matrix F, specifically:

[0106]

[0107] The tag information of other non-black-start power sources and load nodes in the power grid needs to be updated and obtained during the propagation process. Therefore, except for the elements involved in the above formula, all other elements are set to zero.

[0108] Traditional label propagation algorithms primarily rely on the edge weights and label information of neighboring nodes during label updates. Their label propagation strategy is as follows:

[0109]

[0110] In the formula, f(i) represents the label information of node i, C represents the type of label information, N(i) represents the set of adjacent nodes of node i, and w i (f j ) indicates that the target node i and all label information are f j The sum of the edge weights of the adjacent nodes;

[0111] The propagation probability matrix T measures the probability of label information propagating between nodes, and is determined by the weight of the edges. The specific calculation method is as follows:

[0112]

[0113] In the formula, T ij This represents the probability that the label information of node j will propagate to node i.

[0114] To address this problem, the traditional label propagation algorithm is improved by introducing a partition size matrix into the propagation strategy for dividing the power system recovery partitions.

[0115] Define the partition size factor as follows:

[0116]

[0117] In the formula, m c The partition size factor for label category c, d c This represents the sum of the degrees of all nodes with label category c, where l is the total number of branches in the power grid;

[0118] The diagonal elements of the partition size matrix M are the corresponding partition size factor values, specifically:

[0119]

[0120] The M matrix represents the size of existing partitions in the system. The larger the partition size of the label, the smaller the corresponding size influence factor.

[0121] Before each round of label propagation, the current power grid partition size matrix is ​​calculated and incorporated into the considerations for node label updates;

[0122] The improved label propagation strategy is as follows:

[0123]

[0124] In the formula, f(i) represents the label information of node i, C represents the type of label information, N(i) represents the set of adjacent nodes of node i, and w i (f j ) indicates that the target node i and all label information are f j The sum of the edge weights of the adjacent nodes, m(f)j ) indicates that the tag information is f j Community size is an influencing factor.

[0125] Each node propagates according to the propagation probability and social scale matrix, specifically as follows:

[0126] Y t =F t-1 M

[0127] In the formula, the Y matrix is ​​the label influence matrix, which represents the impact of this round of propagation on the label updates of each node;

[0128] Update the label classification matrix F for this round based on the label influence matrix Y. t If the element in the c-th column of the i-th row has the largest value, then the corresponding label classification matrix F... ic The element is set to 1, and the other elements in the i-th row of the F matrix are set to 0, thereby updating the label information of each node;

[0129] The propagation process is repeated until the label classification matrix F converges, yielding the final label information and partitioning results for each node.

[0130] Furthermore, the recovery path search for the unit to be started described in step 2 is as follows:

[0131] The adjacency matrix A is used to characterize the connectivity of the power grid topology, specifically as follows:

[0132]

[0133]

[0134] Define set V S Store the nodes where the shortest path has been found, V o Store other nodes, set D stores the distances between the black start power node and each node, matrix S stores the shortest paths between each node, and set V... S Initially, it only contained a black start power node;

[0135] According to the adjacency matrix, in V o Select the node k with the shortest distance to the black start power source and add it to set V. S Record the shortest path in matrix S, and simultaneously remove the node from set V. o Remove from;

[0136] Compare the path length obtained with node k as the intermediate node with the original path length, take the minimum of the two, and update and adjust the distance from each node in set D to the starting point.

[0137] Repeat the above steps until all nodes have been traversed. This will give you the shortest path and shortest distance from all nodes to the black start power node. Use the shortest path as the recovery path for the node to be started.

[0138] The comprehensive indicators for constructing the unit startup sequence described in step 2 are as follows:

[0139] Taking into account the unit's starting power, ramp rate, and capacity, the specific characteristics of the computer group are as follows:

[0140] O(k)=P * (k)-S * (k)-C * (k)

[0141] In the formula, O(k) is the characteristic index of the k-th unit, and P * (k) is the normalized value of the starting power of the k-th unit, S * (k) is the normalized value of the capacity of the k-th unit, C * (k) is the normalized value of the ramp rate of the k-th unit.

[0142] The normalization method is as follows:

[0143]

[0144] In the formula, x * Represents the normalized value of x, x min Represents the minimum value of x, x max This represents the maximum value of x.

[0145] Based on the recovery path of the unit node to be started, the sum of the branch weights contained in the recovery path is calculated and normalized as the unit distance index D(k).

[0146] Taking into account both the unit's own characteristic indicators and distance indicators, a comprehensive unit startup index is established, specifically as follows:

[0147] Z(k) = O(k) + D(k)

[0148] In the formula, Z(k) is the comprehensive startup index of the k-th unit, O(k) is the characteristic index of the k-th unit, and D(k) is the distance index of the k-th unit.

[0149] Furthermore, the objective function for the multi-black start power supply deployment planning in step 3 is as follows:

[0150] The multi-black-start power supply deployment scheme aims to minimize the black-start recovery time of units with a backup power grid. The optimization objective function can be expressed as:

[0151] minT grid =max(T) cc∈[1,2,…,k]

[0152] Among them, T grid Indicates the unit's black start recovery time; k is the number of black start power supplies, equal to the number of recovery partitions; T c Indicates the black boot time of partition C:

[0153]

[0154] In the formula, a represents the set of black-boot units within partition c, b represents the set of non-black-boot units within partition c, and t m i,BSU t represents the recovery time of the i-th black start unit. m j,NBSU This indicates the recovery time of the j-th black starter unit.

[0155] The specific constraints for the multi-black start power supply deployment planning in step 3 are as follows:

[0156] The installation location of the black starter unit should be within the node range of the power system, specifically:

[0157]

[0158] In the formula, s black,i N represents the location of the i-th black starter unit. bus This represents the set of nodes in the power grid, where n represents the number of nodes in the power grid.

[0159] The installed capacity of the black starter unit should be less than the specified upper limit, specifically:

[0160] 0 <C black,i ≤C max

[0161] In the formula, C black,i C represents the installed capacity of the i-th black starter unit. max This represents the upper limit of the capacity of the black start unit;

[0162] The generator set must be started within the unit's hot start time limit in order for the unit to recover quickly using a hot start method. Specifically:

[0163]

[0164] In the formula, t s For non-black start units, t is the start-up time. s max The maximum hot start time limit for non-black start units refers to the maximum power outage time during which the unit can perform a hot start.

[0165] When the unit's startup time exceeds the minimum warm-start time limit, the unit can only perform a cold start after completing a series of preparations within a certain time delay. In this case, the unit's startup time must meet the minimum cold start time limit, specifically:

[0166]

[0167] In the formula, t s For non-black start units, t is the start-up time. s min Minimum cold start time limit for non-black start units;

[0168] Non-black start units rely on external power for startup. A unit can only start successfully if the power output provided by the system is sufficient to meet the startup power requirements of the next unit to be started. Specifically:

[0169]

[0170] In the formula, p represents the number of black-start units in the system. Let represent the active power output by the i-th black-start generator at time t, and q represent the number of non-black-start generators that have been restored in the system. P represents the active power output by the j-th non-black start unit at time t. st This indicates the starting power required for the next unit to be started;

[0171] The optimal recovery path for unit startup can be determined using Dijkstra's algorithm. The startup time of the unit to be started must meet the charging time constraint of the recovery path, specifically:

[0172] t i,st ≥t j,st +t ij

[0173] In the formula, t i,st t represents the start-up time of unit i to be started. j,st Indicates the start-up time of unit j that has been restored, t ij This indicates the charging time of the recovery path from the restored system to the unit i to be started;

[0174] When starting each non-black start generator unit via the recovery path, the generator unit's output power should be within the upper and lower power limits, specifically:

[0175]

[0176] In the formula, P Gi This represents the active power generated by the i-th generator. This represents the minimum active power that the i-th generator is allowed to output. Q represents the maximum active power that the i-th generator is allowed to output; Gi Let be the reactive power generated by the i-th generator set. Let be the minimum reactive power that the i-th generator set is allowed to output. Let be the maximum reactive power that the i-th generator set is allowed to output;

[0177] The active power transmitted through the line should meet the upper power limit requirement, specifically:

[0178]

[0179] In the formula, P Li Let be the power flowing through the i-th line. The upper limit of power transmitted by the i-th line. The active power transmitted by the line must meet the upper limit of power requirement.

[0180] In practical implementation, this embodiment uses the New England IEEE-118 node standard example for simulation verification of the method. The IEEE-118 node system has 54 generator nodes, 64 load nodes, and 186 branches, as shown in the topology diagram below. Figure 1 As shown.

[0181] A flowchart of a multi-black boot power supply deployment planning method considering partition recovery according to the present invention is shown below. Figure 2 As shown, it includes the following steps:

[0182] S1. Abstract the power grid topology for the planned black-start power source and construct a weighted topology map based on branch recovery time and active power. Establish a power system partitioning model based on an improved label propagation algorithm to ensure that the partition size is as balanced as possible, the internal connections within each partition are close, and the inter-partition connections are sparse.

[0183] The construction of the power grid topology diagram described in S1 is specifically as follows:

[0184] Generators and loads in the power grid are extracted as indistinguishable nodes, and lines and transformers are regarded as branches. The power grid is abstracted as a topological graph G = (V, E), where V represents the set of vertices of the graph and E represents the set of branches of the graph.

[0185] Branch weights are set by comprehensively considering the branch recovery time and the active power flowing through each branch before the power outage, thus obtaining a weighted topology. The shorter the branch recovery time, the greater the weight, and the more likely the two nodes connected by the branch are to be assigned to the same partition. At the same time, the power flow data before the power outage directly reflects the tightness of the connections between various nodes in the power system. The larger the active power value transmitted on the branch, the tighter the connection between the two nodes connected by the branch. Specifically:

[0186]

[0187] In the formula, W ij |P represents the branch weight between node i and node j. ij | represents the absolute value of the active power flowing through the branch between node i and node j before the power outage, t ij This represents the recovery time of the branch between node i and node j. * indicates that the value has been normalized. If nodes i and j have a directly connected branch, according to P... ij t ij The size of the value determines its weight; otherwise, the weight is 0.

[0188] The power system partitioning model established based on the improved label propagation algorithm described in S1 is as follows:

[0189] According to the basic requirements for power system restoration zoning, each zone must be equipped with at least one black-start power source, and it is assumed that the number of zones is consistent with the number of black-start power sources. Assume there are n nodes in the power grid and k black-start power sources (located at x1, x2, ..., xk). k The node containing the black-starting power source is used as the initial labeled node. Its label information remains unchanged throughout the propagation process. Values ​​are assigned to the initial label category matrix F, specifically:

[0190]

[0191] The tag information of other non-black-start power sources and load nodes in the power grid needs to be updated and obtained during the propagation process. Therefore, except for the elements involved in the above formula, all other elements are set to zero.

[0192] Traditional label propagation algorithms primarily rely on the edge weights and label information of neighboring nodes during label updates. Their label propagation strategy is as follows:

[0193]

[0194] In the formula, f(i) represents the label information of node i, C represents the type of label information, N(i) represents the set of adjacent nodes of node i, and w i (f j ) indicates that the target node i and all label information are f j The sum of the edge weights of the adjacent nodes;

[0195] The propagation probability matrix T measures the probability of label information propagating between nodes, and is determined by the weight of the edges. The specific calculation method is as follows:

[0196]

[0197] In the formula, T ij This represents the probability that the label information of node j will propagate to node i.

[0198] To address this problem, the traditional label propagation algorithm is improved by introducing a partition size matrix into the propagation strategy for dividing the power system recovery partitions.

[0199] Define the partition size factor as follows:

[0200]

[0201] In the formula, m c The partition size factor for label category c, d c This represents the sum of the degrees of all nodes with label category c, where l is the total number of branches in the power grid;

[0202] The diagonal elements of the partition size matrix M are the corresponding partition size factor values, specifically:

[0203]

[0204] The M matrix represents the size of existing partitions in the system. The larger the partition size of the label, the smaller the corresponding size influence factor.

[0205] Before each round of label propagation, the current power grid partition size matrix is ​​calculated and incorporated into the considerations for node label updates;

[0206] The improved label propagation strategy is as follows:

[0207]

[0208] In the formula, f(i) represents the label information of node i, C represents the type of label information, N(i) represents the set of adjacent nodes of node i, and w i (f j ) indicates that the target node i and all label information are f j The sum of the edge weights of the adjacent nodes, m(f) j ) indicates that the tag information is f j Community size is an influencing factor.

[0209] Each node propagates according to the propagation probability and social scale matrix, specifically as follows:

[0210] Y t =F t-1 M

[0211] In the formula, the Y matrix is ​​the label influence matrix, which represents the impact of this round of propagation on the label updates of each node;

[0212] Update the label classification matrix F for this round based on the label influence matrix Y. t If the element in the c-th column of the i-th row has the largest value, then the corresponding label classification matrix F... icThe element is set to 1, and the other elements in the i-th row of the F matrix are set to 0, thereby updating the label information of each node;

[0213] The propagation process is repeated until the label classification matrix F converges, yielding the final label information and partitioning results for each node.

[0214] S2. Based on the partitioning results obtained in S1, the recovery path of the units to be started in each partition is searched using the Dijkstra algorithm. Combining the characteristics of unit capacity, ramp rate, start-up power, and the distance of the recovery path, a comprehensive index of unit start-up sequence is constructed, and a unit partition recovery model is established.

[0215] The specific steps for searching the recovery path of the unit to be started, as described in S2, are as follows:

[0216] The adjacency matrix A is used to characterize the connectivity of the power grid topology, specifically as follows:

[0217]

[0218]

[0219] Define set V S Store the nodes where the shortest path has been found, V o Store other nodes, set D stores the distances between the black start power node and each node, matrix S stores the shortest paths between each node, and set V... S Initially, it only contained a black start power node;

[0220] According to the adjacency matrix, in V o Select the node k with the shortest distance to the black start power source and add it to set V. S Record the shortest path in matrix S, and simultaneously remove the node from set V. o Remove from;

[0221] Compare the path length obtained with node k as the intermediate node with the original path length, take the minimum of the two, and update and adjust the distance from each node in set D to the starting point.

[0222] Repeat the above steps until all nodes are traversed. This will give you the shortest path and shortest distance from all nodes to the black start power node. Use the shortest path as the recovery path for the node to be started.

[0223] The specific comprehensive indicators for the unit startup sequence described in S2 are as follows:

[0224] Taking into account the unit's starting power, ramp rate, and capacity, the specific characteristics of the computer group are as follows:

[0225] O(k)=P * (k)-S* (k)-C * (k)

[0226] In the formula, O(k) is the characteristic index of the k-th unit, and P * (k) is the normalized value of the starting power of the k-th unit, S * (k) is the normalized value of the capacity of the k-th unit, C * (k) is the normalized value of the ramp rate of the k-th unit.

[0227] The normalization method is as follows:

[0228]

[0229] In the formula, x * Represents the normalized value of x, x min Represents the minimum value of x, x max This represents the maximum value of x.

[0230] Based on the recovery path of the unit node to be started, the sum of the branch weights contained in the recovery path is calculated and normalized as the unit distance index D(k).

[0231] Taking into account both the unit's own characteristic indicators and distance indicators, a comprehensive unit startup index is established, specifically as follows:

[0232] Z(k) = O(k) + D(k)

[0233] In the formula, Z(k) is the comprehensive startup index of the k-th unit, O(k) is the characteristic index of the k-th unit, and D(k) is the distance index of the k-th unit.

[0234] S3. Based on the unit zoning recovery model obtained in S2, establish a multi-black start power supply layout planning model that considers zoning recovery. It includes two parts: objective function and constraints. The objective function is to minimize the unit black start recovery time, and the constraints include the location of black start units, the capacity of black start units, the unit start-up time, the unit start-up power, and the path charging time.

[0235] The objective function for the multi-black start-up power supply deployment planning described in S3 is as follows:

[0236] The multi-black-start power supply deployment scheme aims to minimize the black-start recovery time of units with a backup power grid. The optimization objective function can be expressed as:

[0237] minT grid =max(T) c c∈[1,2,…,k]

[0238] Among them, T gridIndicates the unit's black start recovery time; k is the number of black start power supplies, equal to the number of recovery partitions; T c Indicates the black boot time of partition C:

[0239]

[0240] In the formula, a represents the set of black-boot units within partition c, b represents the set of non-black-boot units within partition c, and t m i,BSU t represents the recovery time of the i-th black-start unit. m j,NBSU This indicates the recovery time of the j-th black starter unit.

[0241] The specific constraints for the multi-black start-up power supply deployment planning described in S3 are as follows:

[0242] The installation location of the black starter unit should be within the node range of the power system, specifically:

[0243]

[0244] In the formula, s black,i N represents the location of the i-th black starter unit. bus This represents the set of nodes in the power grid, where n represents the number of nodes in the power grid.

[0245] The installed capacity of the black starter unit should be less than the specified upper limit, specifically:

[0246] 0 <C black,i ≤C max

[0247] In the formula, C black,i C represents the installed capacity of the i-th black starter unit. max This represents the upper limit of the capacity of the black start unit;

[0248] The generator set must be started within the unit's hot start time limit in order for the unit to recover quickly using a hot start method. Specifically:

[0249]

[0250] In the formula, t s For non-black start units, t is the start-up time. s max The maximum hot start time limit for non-black start units refers to the maximum power outage time during which the unit can perform a hot start.

[0251] When the unit's startup time exceeds the minimum warm-start time limit, the unit can only perform a cold start after completing a series of preparations within a certain time delay. In this case, the unit's startup time must meet the minimum cold start time limit, specifically:

[0252]

[0253] In the formula, t s For non-black start units, t is the start-up time. s min Minimum cold start time limit for non-black start units;

[0254] Non-black start units rely on external power for startup. A unit can only start successfully if the power output provided by the system is sufficient to meet the startup power requirements of the next unit to be started. Specifically:

[0255]

[0256] In the formula, p represents the number of black-start generator units in the system. Let represent the active power output by the i-th black-start generator at time t, and q represent the number of non-black-start generators that have been restored in the system. P represents the active power output by the j-th non-black start unit at time t. st This indicates the starting power required for the next unit to be started;

[0257] The optimal recovery path for unit startup can be obtained using Dijkstra's algorithm. The startup time of the unit to be started must meet the charging time constraint of the recovery path, specifically:

[0258] t i,st ≥t j,st +t ij

[0259] In the formula, t i,st t represents the start-up time of unit i to be started. j,st Indicates the start-up time of unit j that has been restored, t ij This indicates the charging time of the recovery path from the restored system to the unit i to be started;

[0260] When starting each non-black start generator unit via the recovery path, the generator unit's output power should be within the upper and lower power limits, specifically:

[0261]

[0262] In the formula, P Gi This represents the active power generated by the i-th generator. This represents the minimum active power that the i-th generator is allowed to output. Q represents the maximum active power that the i-th generator is allowed to output; Gi Let be the reactive power generated by the i-th generator set. Let be the minimum reactive power that the i-th generator set is allowed to output. Let be the maximum reactive power that the i-th generator set is allowed to output;

[0263] The active power transmitted through the line should meet the upper power limit requirement, specifically:

[0264]

[0265] In the formula, P Li Let be the power flowing through the i-th line. The upper limit of power transmitted by the i-th line. The active power transmitted by the line must meet the upper limit of power requirement.

[0266] S4. Use an intelligent optimization algorithm to solve the multi-black boot power supply layout planning model described in S3, and obtain a multi-black boot power supply layout planning scheme that considers partition recovery.

[0267] Assuming the new black-start power source is an FCB unit, which can be installed at any node of the generator, with an upper limit of 1000MW for installed capacity. Consider installing 1, 2, and 3 FCB units as black-start power sources. The black-start power source deployment schemes obtained based on the intelligent optimization algorithm are shown in Table 1. The deployment scheme for installing 3 FCB units is as follows: Figure 3 As shown.

[0268] Table 1 Comparison of the number of black start power supplies

[0269]

[0270] If only one black-start generator unit is installed in the 118-node system, the system recovery time is 162.49 minutes after optimization. If two black-start generator units are considered, the optimal installation locations are at nodes 40 and 105, and the corresponding recovery time is 130.37 minutes. If three black-start generator units are added to the system, according to the optimization results, the system recovery time is shortened to 108.17 minutes when the three units are installed at nodes 19, 56, and 100 respectively, which is a reduction of 33.43% compared to installing only one black-start generator. Increasing the number of black-start generators allows for more efficient parallel recovery of multiple zones during the power system restoration process, thereby significantly shortening the total recovery time. Therefore, the duration of power outages can be controlled by adding black-start generators.

[0271] The above are merely preferred embodiments of the present invention and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should recognize that any equivalent substitutions and obvious changes made based on the content of this specification should be included within the protection scope of the present invention.

Claims

1. A method for planning the deployment of multi-black boot power supplies considering partition recovery, characterized in that, include: Abstract the power grid topology of the black-start power source to be planned and constructed, and construct a weighted topology diagram of the power grid based on the branch recovery time and active power. A power system partitioning model was established based on an improved label propagation algorithm. The establishment of the power system partitioning model based on the improved label propagation algorithm includes: According to the basic requirements for power system restoration zoning, each zone must be equipped with at least one black-start power supply, and the number of zones must be consistent with the number of black-start power supplies; assuming there are... n 1 node k Taiwan black starter power supply, respectively located in x 1, x 2,…, x k The node containing the black-start power source is taken as the initial labeled node, and its label information remains unchanged throughout the propagation process, forming the initial stage label category matrix. F Assignment; The tagging information of other non-black-start power sources and load nodes in the power grid needs to be updated and obtained during the propagation process; The traditional label propagation algorithm is improved by introducing a partition size matrix into the propagation strategy for dividing the power system recovery partitions; Define the partition size factor as follows: In the formula, m c Indicates label category c partition size factor, d c Indicates that the tag category is c The sum of the degrees of all nodes l This represents the total number of branches in the power grid. Partition Scale Matrix M The diagonal elements are the corresponding partition size factor values, specifically: M In the matrix representation system, the size of the existing partitions is such that the larger the partition size of the label, the smaller the corresponding size influence factor. Based on the power system zoning model, the recovery path of the units to be started in each zone is searched using the Dijkstra algorithm. Combining the unit capacity, ramp rate, starting power characteristics and the distance of the recovery path, a comprehensive index of unit start-up sequence is constructed, and a zoning recovery model of the units is established. Based on the unit zonal recovery model, a multi-black start power supply deployment planning model for zonal recovery is established, which includes two parts: objective function and constraints. The objective function is to minimize the unit black start recovery time, and the constraints include the location of the black start unit, the capacity of the black start unit, the unit start time, the unit start power, and the path charging time. An intelligent optimization algorithm is used to solve the multi-black boot power supply layout planning model, resulting in a multi-black boot power supply layout planning scheme that considers partition recovery.

2. The multi-black boot power supply deployment planning method considering partition recovery according to claim 1, characterized in that: The construction of the power grid weighted topology map includes: Generators and loads in the power grid are extracted as indistinguishable nodes, and lines and transformers are treated as branches, thus abstracting the power grid into a topology graph. , where V represents the set of vertices of the graph and E represents the set of branches of the graph; The recovery time of each branch and the active power flowing through each branch before the power outage are set as branch weights to obtain a weighted topology, specifically: In the formula, This represents the branch weight between node i and node j. This represents the absolute value of the active power flowing through the branch between node i and node j before the power outage. This represents the recovery time of the branch between node i and node j. This indicates that the numerical values ​​have been normalized. If nodes i and j have directly connected branches, according to , The size of the value determines its weight; otherwise, the weight is 0.

3. The multi-black boot power supply deployment planning method considering partition recovery according to claim 1, characterized in that: The power system partitioning model based on the improved label propagation algorithm also includes: Before each round of label propagation, the current power grid partition size matrix is ​​calculated and incorporated into the considerations for node label updates; The improved tag propagation method is as follows: In the formula, Represents a node i Tag information, C This represents the type of label information. Representative node i The set of adjacent nodes, Represents the target node i With all tag information The sum of the edge weights of the adjacent nodes. The label information is Community size influencing factors; Each node propagates according to the propagation probability and social scale matrix, specifically as follows: In the formula Y The matrix is ​​the label influence matrix, representing the impact of this round of propagation on the label updates of each node; According to the label influence matrix Y Update the label classification matrix for this round. F ,like Y t The i In the middle of the line c If the column element has the largest value, then the corresponding label in the classification matrix will be included. F ic Set the element to 1, and at the same time F Matrix number i Set all other elements in the row to 0 to update the label information of each node; The propagation process is repeated continuously until the label classification matrix is ​​formed. F Convergence yields the final label information and partitioning results for each node.

4. The multi-black boot power supply deployment planning method considering partition recovery according to claim 1, characterized in that, The recovery path for the bootable unit in each partition, searched using the Dijkstra algorithm, includes: The connection relationships of the power grid topology are characterized by the adjacency matrix A as follows: Define a set V S Store the nodes where the shortest path has been found. V o Store other nodes, collection D Store the distances between the black-start power node and each node in the matrix. S Store the shortest paths for each node, in a set V S Initially, it only contained a black start power node; According to the adjacency matrix, V o Select the node with the shortest distance to the black start power supply. k To include it in the set V S Record the shortest path to S Matrix, and simultaneously remove the node from the set V o Remove from; Will be based on nodes k The path length obtained from the intermediate node is compared with the original path length, and the minimum value between the two is used to update and adjust the set. D The distance from each node to the starting point; Traverse all nodes to obtain the shortest path and shortest distance from all nodes to the black start power node, and use the shortest path as the recovery path for the node to be started.

5. The multi-black boot power supply deployment planning method considering partition recovery according to claim 1, characterized in that, The comprehensive indicators for the startup sequence of the construction units include: Based on the unit's starting power, ramp rate, and unit capacity, the following computer group characteristic indicators are calculated: In the formula, For the characteristic indicators of the k-th unit, Let be the normalized value of the starting power of the k-th unit. Let be the normalized value of the capacity of the k-th unit. Let be the normalized value of the ramp rate of the k-th unit; Normalization method: In the formula, represent x The normalized value, Represents the minimum value of x. This represents the maximum value of x; Based on the recovery path of the unit node to be started, the sum of the branch weights contained in the recovery path is calculated and normalized as the unit distance index. ; Based on the unit's own characteristic indicators and distance indicators, establish comprehensive unit startup indicators: In the formula, For the first k Overall startup performance indicators for the Taiwanese generator unit For the first k The performance characteristics of the Taiwanese generator set For the first k Distance indicators for the Taiwanese unit.

6. The multi-black boot power supply deployment planning method considering partition recovery according to claim 1, characterized in that, The objective function of the multi-black boot power supply deployment planning model for partition recovery includes: The multi-black-start power supply deployment scheme aims to minimize the black-start recovery time of units with a backup power grid. The optimization objective function can be expressed as: in, T grid Indicates the unit's black start recovery time; k The number of black boot power supplies is equal to the number of recovery partitions; T c Indicates partition c Black boot time: In the formula, a Indicates partition c Collection of internal black starter units, b Indicates partition c Collection of non-black starter units, Indicates the first i Recovery time for a black start unit Indicates the first j Recovery time for the blackout starter unit.

7. The multi-black boot power supply deployment planning method considering partition recovery according to claim 1, characterized in that, The constraints of the multi-black boot power supply deployment planning model for partitioned recovery include: The installation location of the black starter unit is within the node range of the power system, specifically: In the formula, Indicates the first i Location of each black starter unit Represents the set of nodes in the power grid. n Represents the number of nodes in the power grid; The installed capacity of the black starter unit is less than the specified upper limit, specifically: In the formula, Representing the i The installed capacity of the black starter unit This represents the upper limit of the capacity of the black start unit; The generator set starts within the unit's hot start time limit, and the unit resumes operation using a hot start method, specifically: In the formula, t s The start-up time for non-black start units. t s max This is the maximum hot start time limit for non-black start units, i.e., the maximum power outage time for the unit to perform a hot start. When the unit's startup time exceeds the minimum warm-start time limit, the unit will undergo a cold start. The unit's startup time must meet the minimum cold start time limit, specifically: In the formula, t s The start-up time for non-black start units. t s min Minimum cold start time limit for non-black start units; Non-black start units rely on external power for startup. The unit starts when the system's power output meets the startup power requirement of the next unit to be started. Specifically: In the formula, p This indicates the number of black-start units in the system. express t Time of the first i The amount of active power output by the generator set in Taiwan. q This indicates the number of non-black-start units that have been restored in the system. express t Time of the first j The amount of active power output by the non-black starter unit. This indicates the starting power required for the next unit to be started; The optimal recovery path for unit startup is obtained using Dijkstra's algorithm, and the startup time of the unit to be started satisfies the recovery path charging time constraint: In the formula, Indicates the unit to be started i At the start time, This indicates that the unit has been restored. j At the start time, This indicates a transfer from the restored system to the unit to be started. i The power restoration path charging time; When each non-black start generator unit is started via the recovery path, the power output of the generator unit is within the upper and lower power limits, specifically: In the formula, Representing the The active power generated by the generator, Representing the The minimum active power output allowed by the generator. Representing the The maximum active power output allowed by the generator; For the first The reactive power generated by the generator set For the first The minimum reactive power that a generator set is allowed to output. For the first The maximum reactive power that a generator set is allowed to output; The active power transmitted through the line should meet the upper power limit requirement, specifically: In the formula, P Li For the first i The power flowing through the line, No. i The upper limit of power transmitted by each line; the active power transmitted by the line meets the upper limit of power requirements.

Images