A power supply method and computer-readable medium for optimizing black start power distribution

Abstract

The present invention proposes a power supply method and computer-readable medium for optimizing the placement of black start power sources. The present invention uses minimizing the black start recovery time of non-black start power source nodes as the objective function, optimizes the selection of multiple black start power source nodes from multiple black start power source nodes as restoration power supply nodes, and uses the capacity of the restored black start power source nodes as decision variables to construct a multi-black start power source placement optimization model with minimizing the black start recovery time as the objective. The multi-black start power source placement optimization model is solved based on the Coyote optimization algorithm. The Coyote optimization algorithm is improved by improving the Coyote growth model and introducing adaptive Levy flight and chaotic perturbation mechanisms to obtain the optimal placement location capacity for power restoration in the distribution network. The present invention has stronger global optimization capabilities, high algorithm accuracy, and strong stability. It can distribute multiple black start power sources more evenly across a topological network, shortening the power system recovery time.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system restoration, and in particular relates to a power supply method for optimizing black start power distribution and a computer-readable medium. Background Art

[0002] In recent years, widespread power outages have continued to occur due to factors such as human error, power equipment failure, and extreme weather events. Black-start power supply, the foundation for power system restoration, has become a key factor limiting grid restoration efficiency. The deployment of black-start power supply directly impacts the effectiveness of power system restoration.

[0003] Currently, most research on black-start power generation placement solutions relies on manual methods, and optimization solutions need to be improved. For example, recursive methods are used to solve the optimal black-start unit placement plan, but they are not adaptable enough for large-scale power grids with a large number of units and loads. Ordinal optimization theory is used to simplify complex power system restoration models and solve the optimal FCB unit placement solution, ensuring that optimization efficiency is improved while meeting actual project requirements. The generalized ideal point method is used for solution, but it requires prioritizing the optimal solution set range.

[0004] Existing optimization algorithms can be categorized into two types: traditional optimization algorithms and artificial intelligence algorithms. Traditional algorithms, such as linear programming, nonlinear programming, and interior-point methods, all randomly select an initial point. Starting from this point, guided by the derivatives of the objective function and constraints, they move through the solution space to find the optimal point. Therefore, traditional optimization algorithms require the objective function and constraints to be differentiable, making them incapable of solving problems with a mixture of continuous and discrete variables. They are highly dependent on the initial solution and prone to falling into local optima. Artificial intelligence algorithms, on the other hand, have strong global optimization capabilities and perform better in solving optimization problems with complex fitness functions and constraints, and with a mixture of continuous and discrete variables. Current research has applied artificial intelligence algorithms to solve black-start power supply location optimization problems, such as using genetic algorithms to solve black-start power supply planning models and using particle swarm optimization algorithms to search for optimal location solutions. However, the accuracy and stability of the solutions remain to be improved. Numerous intelligent optimization algorithms exist, and it is crucial to select the appropriate algorithm for solving black-start power supply location optimization problems to improve both accuracy and efficiency. Summary of the Invention

[0005] In view of the deficiencies in the prior art, the present invention proposes a power supply method for optimizing black start power distribution and a computer-readable medium.

[0006] The technical solution of the method of the present invention is a power supply method for optimizing the layout of black start power supplies, comprising the following steps:

[0007] Step 1: Construct a multi-power node distribution network model. In the multi-power node distribution network model, select multiple power nodes as multiple black start power nodes and multiple non-black start power nodes. Use the Dijkstra algorithm to obtain the shortest path between each non-black start power node and each black start power node. Calculate the characteristic index and distance index of each non-black start power node, and calculate the comprehensive index of each non-black start power node.

[0008] Step 2: Construct black start power supply node layout planning constraints. Minimizing the black start recovery time of non-black start power supply nodes is used as the objective function. Optimize the selection of multiple black start power supply nodes from multiple black start power supply nodes as restoration power supply nodes, and use the capacity of the restoration black start power supply nodes as decision variables.

[0009] Step 3: Improve the growth pattern of individual coyotes in the coyote optimization algorithm, introduce the global optimal coyote influence factor to guide the growth direction of individual coyotes, and perform adaptive Lévy flight on the grown coyotes;

[0010] Step 4: Introduce a chaotic optimization perturbation mechanism into the coyote optimization algorithm. Use the Logistic chaotic map to perform a chaotic search for the current global optimal coyote individual. Evaluate the adaptability of all coyote individuals in the resulting chaotic sequence, and use the coyote with the best adaptability to replace the coyote with the worst adaptability.

[0011] Step 5: Optimize and solve the objective function of step 2 through steps 3 and 4 to obtain multiple optimally selected black-start power supply nodes for power restoration and the capacities of multiple optimally selected black-start power supply nodes for power restoration, and obtain multiple optimally selected black-start power supply nodes for power restoration and the capacities of multiple optimally selected black-start power supply nodes for power restoration, which are used for power restoration of the multi-power node distribution network model.

[0012] Preferably, the characteristic index of each non-black start power supply node is calculated in step 1 as follows:

[0013] O(k)=P * (k)-S * (k)-C * (k)

[0014] k∈[1,n G ]

[0015] Where O(k) is the characteristic index of the kth non-black start power node, P * (k) is the normalized value of the startup power of the kth non-black start power node, S * (k) is the normalized value of the capacity of the kth non-black start power node, C *(k) is the normalized value of the ramp rate of the kth non-black start power node, n G Indicates the number of non-black start power nodes;

[0016] The distance index of each non-black start power node is calculated as follows in step 1:

[0017] Among the multiple shortest paths between each non-blackstart power supply node and all start-up power supply nodes, the path with the shortest distance is selected. The distance corresponding to the shortest path is used as the distance index of the non-blackstart power supply node, which is defined as D(k).

[0018] The comprehensive indicators of each non-black start power node are calculated as described in step 1, as follows:

[0019] Z(k)=O(k)+D(k)

[0020] Where Z(k) is the comprehensive startup index of the kth non-blackstart power supply node, O(k) is the characteristic index of the kth non-blackstart power supply node, and D(k) is the distance index of the kth non-blackstart power supply node.

[0021] Preferably, the black start power supply node layout planning constraints in step 2 include:

[0022] Installation location constraints of blackstart power supply nodes, installation capacity constraints of blackstart power supply nodes, startup time constraints of non-blackstart power supply nodes, startup power constraints of blackstart power supply nodes, recovery path charging time constraints of non-blackstart power supply nodes, power constraints of non-blackstart power supply nodes, and branch power constraints of multi-power supply node distribution network models.

[0023] Preferably, the installation position constraint of the black start power supply node is specifically:

[0024] s black,i ∈[1,K]

[0025] i∈[1,p]

[0026] Where p represents the number of black start power supply nodes that have restored power supply, s black,i represents the serial number of the black start power supply node in the multi-power node distribution network model corresponding to the i-th black start power supply node that has restored power supply, and K represents the number of black start power supply nodes in the multi-node distribution network model;

[0027] The installation capacity constraints of the black start power supply node are specifically:

[0028]

[0029] Where, represents the installed capacity of the black start power supply node that has restored power, C max Represents the upper capacity limit of the black start power node;

[0030] The startup time constraint of the non-black start power supply node is specifically:

[0031]

[0032] Where, t s is the start time of the non-black start power node, The maximum hot start time limit for non-black start power nodes;

[0033]

[0034] Where, t s is the start time of the non-black start power node, The minimum cold start time limit for non-black start power nodes;

[0035] The startup power constraint of the black start power supply node is specifically:

[0036]

[0037] Where p represents the number of black start power supply nodes that have restored power supply, It represents the active power outputted by the black start power supply node i that has restored power supply at time t, i.e., it represents the active power outputted by the black start power supply node s at time t. black,i The active power output by the black start power supply node is q, which represents the number of restored non-black start power supply nodes. It represents the active power outputted by the jth restored non-black start power supply node at time t, that is, it represents the active power outputted by the sth restored non-black start power supply node at time t. nblack,j The active power output by a non-black start power node, P st Indicates the starting power required by the next non-black start power node to be started;

[0038] The recovery path charging time constraint of the non-black start power supply node is specifically:

[0039]

[0040] Where, Indicates the startup time of the mth non-black start power node to be restored, that is, the sth oblack,m The startup time of a non-black start power node, represents the startup time of the jth restored non-black start power node, i.e., the sth nblack,j The startup time of a non-black start power node, represents the charging time of the restoration path from the jth restored non-blackstart power node to the mth non-blackstart power node to be restored;

[0041] The power constraints of the non-black start power supply node are specifically:

[0042]

[0043] Where, P G,k represents the active power output of the kth non-black start power node, represents the minimum active power allowed to be output by the i-th non-black start power node, represents the maximum active power allowed to be output by the kth non-black start power node; Q G,k is the reactive power generated by the kth non-black start power node, is the minimum reactive power allowed to be output by the kth non-black start power node, is the maximum reactive power allowed to be output by the kth non-black start power node, n G is the number of non-black start power supply nodes.

[0044] Preferably, the objective function in step 2 is as follows:

[0045] minT grid =T

[0046]

[0047] Where, T grid represents the black start time of the distribution network, T represents the maximum recovery time of all units, t i,black represents the recovery time of the i-th black start power node, t k,nblack It represents the recovery time of the kth non-black start power node, where min represents minimization and max represents maximization.

[0048] Preferably, the growth pattern of the coyote individual in the improved coyote optimization algorithm described in step 3 is as follows:

[0049] Step 3.1: Initialize the coyote population: Set the coyote population size to N and the coyote individual dimension to D. The social state of the coyotes represents the decision variable of the optimization problem. The social state of the cth coyote individual in the pth group during the tth iteration is expressed as:

[0050]

[0051] Where t represents the current iteration number; represents the black start power supply installation location of the c-th individual solution vector in the coyote population during the t-th iteration, represents the black start power supply installed capacity of the c-th individual solution vector in the coyote population during the t-th iteration;

[0052] Divide the coyotes into Np groups, and the number of coyotes in each group is Nc, then N = N p ×N c , initialize the coyote;

[0053]

[0054] Where lb j represents the lower limit of the j-th dimension of the coyote's social state, ub j represents the upper limit of the j-th dimension of the coyote's social state, r j is a random number in [0,1];

[0055] Calculate and evaluate the fitness of individual coyotes:

[0056]

[0057] in, represents the adaptability of the c-th coyote in the p-th group at the t-th iteration, represents the social state of the cth coyote individual in the pth group at the tth iteration, and f(*) represents the objective function with the shortest black start recovery time of the non-black start power supply node;

[0058] Step 3.2: In-group coyote growth: Find the optimal individual in each coyote group:

[0059]

[0060] Among them, alpha p,t represents the leader of the p-th group at the t-th iteration, represents the social state of the c-th coyote in group p at the t-th iteration, Indicates the variable value when the objective function takes the minimum value;

[0061] Arrange the coyotes in the group according to their fitness and get O p,t Sequence, the median of the sequence is used to evaluate the group culture trend of the coyote group, representing the average social situation of the coyote group:

[0062]

[0063] in, represents the cultural trend of the coyote group at the tth iteration of group p, Indicates that when N c When is an odd number, the median of the j-th dimension of all coyotes in the p-th group at the t-th iteration, Indicates that when Nc When is an even number, it is the median of the jth dimension of all coyotes in group p at iteration t;

[0064] Use the best individuals and cultural trends within the group to guide the growth of individual coyotes within the group:

[0065]

[0066]

[0067]

[0068] Where alpha p,t represents the optimal individual coyote in group p at the tth iteration, cult p,t represents the cultural trend within group p at iteration t, and are two coyotes randomly selected from group p at the tth iteration. r1 and r2 represent the first and second growth control parameters, respectively, and are random numbers in [0,1]. δ1 represents the optimal individual influence factor within the group, and δ2 represents the cultural trend influence factor within the group. represents the social state of the cth coyote in group p at the tth iteration after the coyote grows up;

[0069] The social adaptability of coyotes before and after growth is evaluated based on the fitness function, and the social status of coyotes with better adaptability is retained;

[0070]

[0071]

[0072] in, represents the adaptability of the c-th coyote in group p after it grows up at the t-th iteration. represents the social state of the c-th coyote in group p at the t-th iteration after the coyote grows up, and soc represents the social state of the c-th coyote in group p with better adaptability at the t-th iteration;

[0073] The global optimal coyote influence factor δ3 is added to the growth model of the coyote optimization algorithm. This factor can guide the individuals in the group to grow in the direction of the global optimal coyote so far, thereby improving the global optimization ability and convergence speed of the algorithm. The improved growth method of the coyote group is:

[0074]

[0075] Where alpha p,t represents the optimal individual coyote in group p at the tth iteration, cultp,t represents the cultural trend of group p at iteration t, galpha p,t represents the global optimal coyote of group p at iteration t, and are three coyotes randomly selected from group p at the tth iteration, r1, r2, and r3 represent the first, second, and third growth control parameters, respectively, and are all random numbers in [0,1];

[0076] Step 3.3: Coyote birth and death: The birth of a new coyote is the result of the union of two randomly selected parent coyotes within the group, plus the combined effects of environmental factors;

[0077]

[0078] Where, represents the newly born coyotes in the jth dimension of the pth group at the tth iteration, and They represent the two parent coyotes randomly selected from the jth dimension in the pth group at the tth iteration, rnd j is a random number in [0,1], R j is the randomly generated variation value within the j-th dimension of the coyote social state;

[0079] P s =1 / D

[0080] P a =(1-P s ) / 2

[0081] Where, P s is the inheritance probability, P a represents the probability of mutation, and D represents the individual dimension of the coyote;

[0082] The Mantegna algorithm is used to generate random step sizes that obey the Levy distribution, specifically:

[0083]

[0084] Where s represents the random step size of the Levy distribution, γ represents the Levy parameter, μ represents the random number of the Gaussian distribution, and ν represents the random number of the Gaussian distribution.

[0085]

[0086] Among them, σ μ represents the variance of random number μ, σ v represents the variance of the random number ν, Γ represents the gamma function, which is the extension of the factorial function on real and complex numbers, and NO represents the normal distribution;

[0087] The set S of flight steps in each dimension of the social state of coyote individual c c =[s1,s2,…,s D ], the coyote performs Lévy flight with a certain flight probability during the iteration process:

[0088]

[0089] Among them, rnd represents a random number, represents the social state of the cth coyote in group p at iteration t, P l represents the probability of Levy flight, s d Represents the d-th subvector in the flight step set, d∈[1,D];

[0090] Increases Levi's chance of flying, specifically:

[0091]

[0092] Among them, P l represents the probability of Levy flight, P min represents the minimum value of the Levy flight probability, P max represents the maximum value of Levy flight probability, n represents the number of iterations, n max Represents the maximum number of iterations;

[0093] Adopting the elite flight exemption strategy, that is, the current global optimal coyote individual does not participate in the Levy flight and retains its corresponding social state;

[0094] Step 3.4: Coyote Migration: Coyote populations do not remain static after being randomly grouped. Migration sometimes occurs between coyote groups. This ensures diversity within the coyote groups while also allowing for information exchange between different groups.

[0095]

[0096] Where, P e represents the probability that a coyote is driven away by a group and accepted by another group, N c represents the size of coyotes in the group, and ε represents the probability coefficient of coyote migration.

[0097] Preferably, the adaptive Levy flight in step 3 is specifically as follows:

[0098] The Mantegna algorithm is used to generate random step sizes that obey the Levy distribution, specifically:

[0099]

[0100] Among them, γ represents the Levy parameter, μ represents the random number obeying the Gaussian distribution, and ν represents the random number obeying the Gaussian distribution;

[0101]

[0102] Among them, σ μ represents the variance of random number μ, σ v represents the variance of the random number ν, Γ represents the gamma function, which is the extension of the factorial function on real and complex numbers, and NO represents the normal distribution;

[0103] The set S of flight steps in each dimension of the social state of coyote individual c c =[s1,s2,…,s D ], the coyote performs Lévy flight with a certain flight probability during the iteration process:

[0104]

[0105] in, represents the social state of the cth coyote in group p at iteration t, P l represents the probability of Levy flight, s d Represents the d-th subvector in the flight step set, d∈[1,D];

[0106] Increases Levi's chance of flying, specifically:

[0107]

[0108] Among them, P l represents the probability of Levy flight, P min represents the minimum value of the Levy flight probability, P max represents the maximum value of Levy flight probability, n represents the number of iterations, n max Represents the maximum number of iterations;

[0109] An elite flight exemption strategy is adopted, that is, the current global optimal coyote individual does not participate in Levy flight and retains its corresponding social state.

[0110] Preferably, the chaos optimization perturbation mechanism described in step 4 is specifically as follows:

[0111] Set the chaotic mutation probability p h , randomly generate a real number in [0,1], if it is less than p h Then perform chaotic perturbation, otherwise directly enter the next iteration;

[0112] Map each dimension of the global optimal coyote individual at time t to the chaotic space, specifically:

[0113]

[0114] Where, is the j-th dimension chaos value of the optimal coyote in the t-th iteration, is the j-th dimension value of the optimal coyote in the t-th iteration; and are the upper and lower limits of the j-th dimension of the coyote, respectively;

[0115] Perform N-1 Logistic chaotic searches to obtain a chaotic sequence

[0116]

[0117] in, is the j-th chaotic value of the n-th chaotic variable of the global optimal coyote in the t-th iteration, is the j-th chaotic value of the n+1-th chaotic variable of the global optimal coyote in the t-th iteration, is the j-th dimension value of the global optimal coyote in the t-th iteration, represents the upper limit of the j-th dimension of the coyote, represents the lower bound of the j-th dimension of the coyote, n∈[1,N];

[0118] Map the N chaotic variables in the chaotic sequence to the solution space of the optimization problem according to the following formula to generate N new coyote individuals;

[0119]

[0120] in, represents the j-th dimension value of the n-th chaotic variable of the global optimal coyote at the t-th iteration, represents the upper limit of the j-th dimension of the coyote, represents the lower bound of the j-th dimension of the coyote, z n,j represents the j-th chaotic value of the n-th chaotic variable of the global optimal coyote;

[0121] The adaptability of the new coyote individuals is evaluated according to the fitness function, and the coyote individuals with the best adaptability are retained, which is recorded as by The coyote replaces the coyote with the worst adaptability at time t.

[0122] The present invention also provides a computer-readable medium, which stores a computer program executed by an electronic device. When the computer program runs on the electronic device, it enables the steps of the power supply method for optimizing the layout of the black start power supply.

[0123] The present invention has the following advantages:

[0124] The improved coyote optimization algorithm of the present invention has stronger global optimization capability, high algorithm accuracy and strong stability. The proposed black start power supply optimization distribution method can distribute multiple black start power supplies on the topology network in a more balanced manner, shortening the system recovery time. BRIEF DESCRIPTION OF THE DRAWINGS

[0125] Figure 1 : A flow chart of a method according to an embodiment of the present invention;

[0126] Figure 2 : Black start power supply layout result diagram of an embodiment of the present invention;

[0127] Figure 3 : Convergence curve comparison diagram of the embodiment of the present invention;

[0128] Figure 4 : Algorithm stability comparison chart of the embodiments of the present invention. DETAILED DESCRIPTION

[0129] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0130] In specific implementation, the method proposed in the technical solution of the present invention can be automatically run by those skilled in the art using computer software technology. System devices that implement the method, such as computer-readable storage media that store the corresponding computer program of the technical solution of the present invention and computer equipment that runs the corresponding computer program, should also be within the scope of protection of the present invention.

[0131] The present invention uses the New England IEEE-118 node standard example to simulate and verify the method. The IEEE-118 node system has 54 generator nodes, 64 load nodes and 186 branches.

[0132] The following combination Figure 1-4 The technical solution of the method of the embodiment of the present invention is a power supply method for optimizing the layout of black start power supplies, which is specifically as follows:

[0133] The flow chart of the method of the present invention is as follows Figure 1 shown.

[0134] Step 1: Construct a multi-power node distribution network model. In the multi-power node distribution network model, select multiple power nodes as multiple black start power nodes and multiple non-black start power nodes. Use the Dijkstra algorithm to obtain the shortest path between each non-black start power node and each black start power node. Calculate the characteristic index and distance index of each non-black start power node, and calculate the comprehensive index of each non-black start power node.

[0135] The characteristic indicators of each non-black start power node are calculated as described in step 1, as follows:

[0136] O(k)=P * (k)-S * (k)-C * (k)

[0137] k∈[1,n G ]

[0138] Where O(k) is the characteristic index of the kth non-black start power node, P * (k) is the normalized value of the startup power of the kth non-black start power node, S * (k) is the normalized value of the capacity of the kth non-black start power node, C * (k) is the normalized value of the ramp rate of the kth non-black start power node, n G Indicates the number of non-black start power nodes;

[0139] The distance index of each non-black start power node is calculated as follows in step 1:

[0140] Among the multiple shortest paths between each non-blackstart power supply node and all start-up power supply nodes, the path with the shortest distance is selected. The distance corresponding to the shortest path is used as the distance index of the non-blackstart power supply node, which is defined as D(k).

[0141] The comprehensive indicators of each non-black start power node are calculated as described in step 1, as follows:

[0142] Z(k)=O(k)+D(k)

[0143] Where Z(k) is the comprehensive startup index of the kth non-blackstart power supply node, O(k) is the characteristic index of the kth non-blackstart power supply node, and D(k) is the distance index of the kth non-blackstart power supply node;

[0144] Step 2: Construct black start power supply node layout planning constraints. Minimizing the black start recovery time of non-black start power supply nodes is used as the objective function. Optimize the selection of multiple black start power supply nodes from multiple black start power supply nodes as restoration power supply nodes, and use the capacity of the restoration black start power supply nodes as decision variables.

[0145] The black start power supply node layout planning constraints described in step 2 include:

[0146] Installation location constraints of black start power supply nodes, installation capacity constraints of black start power supply nodes, startup time constraints of non-black start power supply nodes, startup power constraints of black start power supply nodes, recovery path charging time constraints of non-black start power supply nodes, power constraints of non-black start power supply nodes, and branch power constraints of multi-power supply node distribution network models;

[0147] The installation position constraints of the black start power supply node are specifically:

[0148] s black,i ∈[1,K]

[0149] i∈[1,p]

[0150] Where p represents the number of black start power supply nodes that have restored power supply, s black,i represents the serial number of the black start power supply node in the multi-power node distribution network model corresponding to the i-th black start power supply node that has restored power supply, and K represents the number of black start power supply nodes in the multi-node distribution network model;

[0151] The installation capacity constraints of the black start power supply node are specifically:

[0152]

[0153] Where, represents the installed capacity of the black start power supply node that has restored power, C max Represents the upper capacity limit of the black start power node;

[0154] The startup time constraint of the non-black start power supply node is specifically:

[0155]

[0156] Where, t s is the start time of the non-black start power node, The maximum hot start time limit for non-black start power nodes;

[0157]

[0158] Where, t s is the start time of the non-black start power node, The minimum cold start time limit for non-black start power nodes;

[0159] The startup power constraint of the black start power supply node is specifically:

[0160]

[0161] Where p represents the number of black start power supply nodes that have restored power supply, It represents the active power outputted by the black start power supply node i that has restored power supply at time t, i.e., it represents the active power outputted by the black start power supply node s at time t. black,i The active power output by the black start power supply node is q, which represents the number of restored non-black start power supply nodes. It represents the active power outputted by the jth restored non-black start power supply node at time t, that is, it represents the active power outputted by the sth restored non-black start power supply node at time t. nblack,j The active power output by a non-black start power node, P st Indicates the starting power required by the next non-black start power node to be started;

[0162] The recovery path charging time constraint of the non-black start power supply node is specifically:

[0163]

[0164] Where, Indicates the startup time of the mth non-black start power node to be restored, that is, the sth oblack,m The startup time of a non-black start power node, represents the startup time of the jth restored non-black start power node, i.e., the sth nblack,j The startup time of a non-black start power node, represents the charging time of the restoration path from the jth restored non-blackstart power node to the mth non-blackstart power node to be restored;

[0165] The power constraints of the non-black start power supply node are specifically:

[0166]

[0167] Where, P G,k represents the active power output of the kth non-black start power node, represents the minimum active power allowed to be output by the i-th non-black start power node, represents the maximum active power allowed to be output by the kth non-black start power node; Q G,k is the reactive power generated by the kth non-black start power node, is the minimum reactive power allowed to be output by the kth non-black start power node, is the maximum reactive power allowed to be output by the kth non-black start power node, n G is the number of non-black start power supply nodes;

[0168] The objective function described in step 2 is as follows:

[0169] minT grid =T

[0170]

[0171] Where, T grid represents the black start time of the distribution network, T represents the maximum recovery time of all units, t i,black represents the recovery time of the i-th black start power supply node, t represents the recovery time of the k-th non-black start power supply node, and min represents the maximum

[0172] k,nblack

[0173] Minimize, and max means maximize.

[0174] Step 3: Improve the growth pattern of individual coyotes in the coyote optimization algorithm, introduce the global optimal coyote influence factor to guide the growth direction of individual coyotes, and perform adaptive Lévy flight on the grown coyotes;

[0175] The growth pattern of coyote individuals in the improved coyote optimization algorithm described in step 3 is as follows:

[0176] Step 3.1: Initialize the coyote population: Set the coyote population size to N and the coyote individual dimension to D. The social state of the coyotes represents the decision variable of the optimization problem. The social state of the cth coyote individual in the pth group during the tth iteration is expressed as:

[0177]

[0178] Where t represents the current iteration number; represents the black start power supply installation location of the c-th individual solution vector in the coyote population during the t-th iteration, represents the black start power supply installed capacity of the c-th individual solution vector in the coyote population during the t-th iteration;

[0179] Divide the coyotes into Np groups, and the number of coyotes in each group is Nc, then N = N p ×N c , initialize the coyote;

[0180]

[0181] Where lb j represents the lower limit of the j-th dimension of the coyote's social state, ub j represents the upper limit of the j-th dimension of the coyote's social state, r j is a random number in [0,1].

[0182] Calculate and evaluate the fitness of individual coyotes:

[0183]

[0184] in, represents the adaptability of the c-th coyote in the p-th group at the t-th iteration, represents the social state of the cth coyote individual in the pth group at the tth iteration, and f(*) represents the objective function with the shortest black start recovery time of the non-black start power supply node;

[0185] Step 3.2: In-group coyote growth: Find the optimal individual in each coyote group:

[0186]

[0187] Among them, alpha p,t represents the leader of the p-th group at the t-th iteration, represents the social state of the c-th coyote in group p at the t-th iteration, Indicates the variable value when the objective function takes the minimum value;

[0188] Arrange the coyotes in the group according to their fitness and get O p,t Sequence, the median of the sequence is used to evaluate the group culture trend of the coyote group, representing the average social situation of the coyote group:

[0189]

[0190] in, represents the cultural trend of the coyote group at the tth iteration of group p, Indicates that when N c When is an odd number, the median of the j-th dimension of all coyotes in the p-th group at the t-th iteration, Indicates that when N c When is an even number, it is the median of the jth dimension of all coyotes in group p at iteration t;

[0191] Use the best individuals and cultural trends within the group to guide the growth of individual coyotes within the group:

[0192]

[0193]

[0194]

[0195] Where alpha p,t represents the optimal individual coyote in group p at the tth iteration, cult p,t represents the cultural trend of group p at iteration t, and are two coyotes randomly selected from group p at the tth iteration. r1 and r2 represent the first and second growth control parameters, respectively, and are random numbers in [0,1]. δ1 represents the optimal individual influence factor within the group, and δ2 represents the cultural trend influence factor within the group. represents the social state of the cth coyote in group p at the tth iteration after the coyote grows up;

[0196] The social adaptability of coyotes before and after growth was evaluated based on the fitness function, and the social states of coyotes with better adaptability were retained.

[0197]

[0198]

[0199] in, represents the adaptability of the c-th coyote in group p after it grows up at the t-th iteration. It represents the social state of the c-th coyote in group p at the t-th iteration after the coyote grows up, and soc represents the social state of the c-th coyote in group p with better adaptability at the t-th iteration.

[0200] The global optimal coyote influence factor δ3 is added to the growth model of the COA algorithm. This factor can guide the individuals in the group to grow towards the global optimal coyote so far, thereby improving the algorithm's global optimization ability and convergence speed. The improved coyote group growth method is:

[0201]

[0202] Where alpha p,t represents the optimal individual coyote in group p at the tth iteration, cult p,t represents the cultural trend of group p at iteration t, galpha p,t represents the global optimal coyote of group p at iteration t, and are three coyotes randomly selected from group p at iteration t, where r1, r2, and r3 are all random numbers in the range [0,1];

[0203] Step 3.3: Coyote birth and death: The birth of a new coyote is the result of the union of two randomly selected parent coyotes within the group, plus the combined effect of environmental factors.

[0204]

[0205] Where, represents the newly born coyotes in the jth dimension of the pth group at the tth iteration, and They represent the two parent coyotes randomly selected from the jth dimension in the pth group at the tth iteration, rnd j is a random number in [0,1], R j is a randomly generated variation value within the j-th dimension of the coyote social state.

[0206] P s =1 / D

[0207] P a =(1-P s ) / 2

[0208] Where, P s is the inheritance probability, P a represents the probability of mutation, and D represents the individual dimension of the coyote.

[0209] The adaptive Levy flight described in step 3 is as follows:

[0210] The Mantegna algorithm is used to generate random step sizes that obey the Levy distribution, specifically:

[0211]

[0212] Among them, γ represents the Levy parameter, μ represents the random number obeying the Gaussian distribution, and ν represents the random number obeying the Gaussian distribution;

[0213]

[0214] Among them, σ μ represents the variance of random number μ, σ v represents the variance of the random number ν, Γ represents the gamma function, which is the extension of the factorial function on real and complex numbers, and NO represents the normal distribution;

[0215] The set S of flight steps in each dimension of the social state of coyote individual c c =[s1,s2,…,s D ], the coyote performs Lévy flight with a certain flight probability during the iteration process:

[0216]

[0217] in, represents the social state of the cth coyote in group p at iteration t, P l represents the probability of Levy flight, s d Represents the d-th subvector in the flight step set, d∈[1,D];

[0218] Increases Levi's chance of flying, specifically:

[0219]

[0220] Among them, P l represents the probability of Levy flight, P min represents the minimum value of the Levy flight probability, P max represents the maximum value of Levy flight probability, n represents the number of iterations, n max Represents the maximum number of iterations;

[0221] Adopting the elite flight exemption strategy, that is, the current global optimal coyote individual does not participate in the Levy flight and retains its corresponding social state;

[0222] Step 3.4: Coyote migration: After the coyote population is randomly grouped, it does not remain fixed. Sometimes migration occurs between coyote groups. On the one hand, it can ensure the diversity of coyote groups, and on the other hand, it can also facilitate information exchange between different groups.

[0223]

[0224] Where, P e represents the probability that a coyote is driven away by a group and accepted by another group, N c represents the size of the coyotes in the group, ε represents the probability coefficient of coyote migration, and ε=0.005 is the preferred value of the present invention.

[0225] The adaptive Levy flight described in step 3 is as follows:

[0226] The Mantegna algorithm is used to generate random step sizes that obey the Levy distribution, specifically:

[0227]

[0228] Among them, γ represents the Levy parameter, μ represents the random number obeying the Gaussian distribution, and ν represents the random number obeying the Gaussian distribution;

[0229]

[0230] Among them, σ μ represents the variance of random number μ, σ v represents the variance of the random number ν, Γ represents the gamma function, which is the extension of the factorial function to real and complex numbers;

[0231] The set S of flight steps in each dimension of the social state of coyote individual c c =[s1,s2,…,s D ], the coyote performs Lévy flight with a certain flight probability during the iteration process:

[0232]

[0233] in, represents the social state of the cth coyote in group p at iteration t, P l represents the probability of Levy flight, s d Represents the d-th subvector in the flight step set, d∈[1,D];

[0234] Increases Levi's chance of flying, specifically:

[0235]

[0236] Among them, P l represents the probability of Levy flight, P min represents the minimum value of the Levy flight probability, P max represents the maximum value of Levy flight probability, n represents the number of iterations, n max Represents the maximum number of iterations;

[0237] Adopting the elite flight exemption strategy, that is, the current global optimal coyote individual does not participate in the Levy flight and retains its corresponding social state;

[0238] Step 4: Introduce a chaotic optimization perturbation mechanism into the coyote optimization algorithm. Use the Logistic chaotic map to perform a chaotic search for the current global optimal coyote individual. Evaluate the adaptability of all coyote individuals in the resulting chaotic sequence, and use the coyote with the best adaptability to replace the coyote with the worst adaptability.

[0239] The chaos optimization perturbation mechanism described in step 4 is as follows:

[0240] Set the chaotic mutation probability p h , randomly generate a real number in [0,1], if it is less than p h Then perform chaotic perturbation, otherwise directly enter the next iteration;

[0241] Map each dimension of the global optimal coyote individual at time t to the chaotic space, specifically:

[0242]

[0243] Where, is the j-th dimension chaos value of the optimal coyote in the t-th iteration, is the j-th dimension value of the optimal coyote in the t-th iteration; and are the upper and lower bounds of the j-th dimension of the coyote, respectively;

[0244] Perform N-1 Logistic chaotic searches to obtain a chaotic sequence

[0245]

[0246] in, is the j-th chaotic value of the n-th chaotic variable of the global optimal coyote in the t-th iteration, is the j-th chaotic value of the n+1-th chaotic variable of the global optimal coyote in the t-th iteration, is the j-th dimension value of the global optimal coyote in the t-th iteration, represents the upper limit of the j-th dimension of the coyote, represents the lower bound of the j-th dimension of the coyote, n∈[1,N];

[0247] Map the N chaotic variables in the chaotic sequence to the solution space of the optimization problem according to the following formula to generate N new coyote individuals;

[0248]

[0249] in, represents the j-th dimension value of the n-th chaotic variable of the global optimal coyote at the t-th iteration, represents the upper limit of the j-th dimension of the coyote, represents the lower bound of the j-th dimension of the coyote, z n,j represents the j-th chaotic value of the n-th chaotic variable of the global optimal coyote;

[0250] The adaptability of the new coyote individuals is evaluated according to the fitness function, and the coyote individuals with the best adaptability are retained, which is recorded as

[0251] by The coyote replaces the coyote individual with the worst adaptability at time t;

[0252] Step 5: Optimize and solve the objective function of step 2 through steps 3 and 4 to obtain multiple optimally selected black-start power supply nodes for power restoration and the capacities of multiple optimally selected black-start power supply nodes for power restoration, and obtain multiple optimally selected black-start power supply nodes for power restoration and the capacities of multiple optimally selected black-start power supply nodes for power restoration, which are used for power supply restoration in the multi-power node distribution network model.

[0253] To verify the effectiveness of the present invention, a comparative analysis is conducted on the black start power supply layout planning methods based on the following three methods:

[0254] Method 1: Particle Swarm Optimization algorithm, defined as PSO algorithm;

[0255] Method 2: Traditional Coyote Optimization Algorithm, defined as COA algorithm;

[0256] Method 3: the improved coyote optimization algorithm of the present invention, defined as the ICOA algorithm;

[0257] The multi-black start power supply layout planning is solved based on three algorithms respectively, and three black start power supply installation schemes are obtained as shown in the table. The black start power supply layout locations are as follows: Figure 2 shown.

[0258] Depend on Figure 3 It can be seen that the black start power supply locations obtained based on methods one and two are concentrated on the right side and below the bottom grid, resulting in a relatively uneven distribution. However, the solution obtained based on method three proposed in this invention can install black start power supplies more evenly on the topological plane, which is conducive to improving the efficiency of parallel power system restoration.

[0259] As shown in the table, the black start power supply placement solution derived from the particle swarm optimization algorithm takes the longest to recover, followed by the solution derived from the Coyote Optimization Algorithm. This demonstrates the superiority of the Coyote Optimization Algorithm in solving the multi-black start power supply placement optimization problem. The black start power supply installation solution derived from the improved Coyote Optimization Algorithm takes only 108.17 minutes to recover, the shortest recovery time. This demonstrates the strong global optimization capability of the proposed method and its effectiveness in shortening power system recovery time and reducing user outage duration.

[0260] Table 1: Comparison of multiple black start power supply installation solutions

[0261]

[0262] The convergence curves of the three methods in the solution process are as follows: Figure 3 As shown, we can see that:

[0263] The particle swarm algorithm has a slow overall search speed and tends to stay at the local optimum. It gradually stabilizes at a recovery time of 116.84 minutes after 57 iterations. The minimum recovery time value obtained is large, and the global optimization ability is poor.

[0264] The coyote optimization algorithm improves the convergence speed to a certain extent. After more than 20 iterations, it performs a local fine search near the convergence point and finally converges after 39 iterations. The recovery time of the convergence point is shortened by 1.93% compared with that of the particle swarm optimization algorithm.

[0265] The improved coyote optimization algorithm proposed in the present invention can quickly approach the optimal point in the early stage of iteration and converge after 26 iterations. It has good optimization speed and global search capability. The time required for the recovery of the black start power distribution scheme is shortened by 7.42% compared with method one and by 5.60% compared with method two.

[0266] In order to further analyze the stability and accuracy of different black start power supply layout methods, 30 independent optimization solutions were performed based on three algorithms. Figure 4 As shown in the table, the average value and mean square error of the recovery time under different algorithms are calculated respectively.

[0267] Table 2: Comparison of multiple black start power supply layout optimization methods

[0268]

[0269] The recovery times corresponding to the placement schemes obtained using methods one and two fluctuated widely, with mean square errors of 1.24 and 1.96, respectively. The algorithmic solution was unstable, and the average recovery time obtained from multiple solutions was relatively large, resulting in poor optimization accuracy. However, the experimental curve corresponding to method three was flat, and the recovery times obtained from multiple experiments were relatively close, with a mean square error of only 0.59 and an average recovery time of 108.70 minutes, the shortest of the three. This demonstrates the effectiveness of optimizing black start power supply placement using the improved Coyote optimization algorithm proposed in this invention in reducing power system recovery time.

[0270] A specific embodiment of the present invention also provides a computer-readable medium.

[0271] The computer readable medium is a server workstation;

[0272] The server workstation stores a computer program executed by an electronic device. When the computer program runs on the electronic device, the electronic device executes the steps of the method for jointly allocating drone trajectories and resources in an embodiment of the present invention.

[0273] It should be understood that parts not elaborated in detail in this specification belong to the prior art.

[0274] It should be understood that the above description of the preferred embodiment is relatively detailed and cannot be regarded as limiting the scope of protection of the patent of the present invention. Under the guidance of the present invention, ordinary technicians in this field can also make substitutions or modifications without departing from the scope of protection of the claims of the present invention, which all fall within the scope of protection of the present invention. The scope of protection requested by the present invention shall be based on the attached claims.

Claims

1. A black start power supply optimization layout method, characterized in that: The following steps are involved: Step 1: Construct a multi-power node distribution network model. In the multi-power node distribution network model, select multiple power nodes as multiple black start power nodes and multiple non-black start power nodes. Use the Dijkstra algorithm to obtain the shortest path between each non-black start power node and each black start power node. Calculate the characteristic index and distance index of each non-black start power node, and calculate the comprehensive index of each non-black start power node. Step 2: Construct black start power supply node layout planning constraints. Minimizing the black start recovery time of non-black start power supply nodes is used as the objective function. Optimize the selection of multiple black start power supply nodes from multiple black start power supply nodes as restoration power supply nodes, and use the capacity of the restoration black start power supply nodes as decision variables. Step 3: Improve the growth pattern of individual coyotes in the coyote optimization algorithm, introduce the global optimal coyote influence factor to guide the growth direction of individual coyotes, and perform adaptive Lévy flight on the grown coyotes; Step 4: Introduce a chaotic optimization perturbation mechanism into the coyote optimization algorithm. Use the Logistic chaotic map to perform a chaotic search for the current global optimal coyote individual. Evaluate the adaptability of all coyote individuals in the resulting chaotic sequence, and use the coyote with the best adaptability to replace the coyote with the worst adaptability. Step 5: Optimize and solve the objective function of step 2 through steps 3 and 4 to obtain multiple optimally selected black-start power supply nodes for power restoration and the capacities of multiple optimally selected black-start power supply nodes for power restoration, and obtain multiple optimally selected black-start power supply nodes for power restoration and the capacities of multiple optimally selected black-start power supply nodes for power restoration, which are used for power restoration in the multi-power node distribution network model; The objective function described in step 2 is as follows: minT grid =T T=max{t i,black |i∈[1,p],t k,nblack |k∈[1,n G ]} Where, T grid represents the black start time of the distribution network, T represents the maximum recovery time of all units, t i,black represents the recovery time of the i-th black start power node, t k,nblack It represents the recovery time of the kth non-black start power node, where min represents minimization and max represents maximization.

2. The power supply method for optimizing black start power distribution according to claim 1, characterized in that: The characteristic indicators of each non-black start power node are calculated as described in step 1, as follows: O(k)=P * (k)-S * (k)-C * (k) k∈[1,n G ] Where O(k) is the characteristic index of the kth non-black start power node, P * (k) is the normalized value of the startup power of the kth non-black start power node, S * (k) is the normalized value of the capacity of the kth non-black start power node, C * (k) is the normalized value of the ramp rate of the kth non-black start power node, n G Indicates the number of non-black start power nodes; The distance index of each non-black start power node is calculated as follows in step 1: Among the multiple shortest paths between each non-blackstart power supply node and all start-up power supply nodes, the path with the shortest distance is selected. The distance corresponding to the shortest path is used as the distance index of the non-blackstart power supply node, which is defined as D(k). The comprehensive indicators of each non-black start power node are calculated as described in step 1, as follows: Z(k)=O(k)+D(k) Where Z(k) is the comprehensive startup index of the kth non-blackstart power supply node, O(k) is the characteristic index of the kth non-blackstart power supply node, and D(k) is the distance index of the kth non-blackstart power supply node.

3. The power supply method for optimizing black start power distribution according to claim 1, characterized in that: The black start power supply node layout planning constraints described in step 2 include: Installation location constraints of blackstart power supply nodes, installation capacity constraints of blackstart power supply nodes, startup time constraints of non-blackstart power supply nodes, startup power constraints of blackstart power supply nodes, recovery path charging time constraints of non-blackstart power supply nodes, power constraints of non-blackstart power supply nodes, and branch power constraints of multi-power supply node distribution network models.

4. The power supply method for optimizing black start power distribution according to claim 3, characterized in that: The installation position constraints of the black start power supply node are specifically: s black,i ∈[1,K] i∈[1,p] Where p represents the number of black start power supply nodes that have restored power supply, s black,i represents the serial number of the black start power supply node in the multi-power node distribution network model corresponding to the i-th black start power supply node that has restored power supply, and K represents the number of black start power supply nodes in the multi-node distribution network model; The installation capacity constraints of the black start power supply node are specifically: Where, represents the installed capacity of the black start power supply node that has restored power, C max Represents the upper capacity limit of the black start power node; The startup time constraint of the non-black start power supply node is specifically: Where, t s is the startup time of the non-black start power node, t s max The maximum hot start time limit for non-black start power nodes; Where, t s is the start time of the non-black start power node, The minimum cold start time limit for non-black start power nodes; The startup power constraint of the black start power supply node is specifically: Where p represents the number of black start power supply nodes that have restored power supply, It represents the active power outputted by the black start power supply node i that has restored power supply at time t, i.e., it represents the active power outputted by the black start power supply node s at time t. black,i The active power output by the black start power supply node is q, which represents the number of restored non-black start power supply nodes. It represents the active power outputted by the jth restored non-black start power supply node at time t, that is, it represents the active power outputted by the sth restored non-black start power supply node at time t. nblack,j The active power output by a non-black start power node, P st Indicates the starting power required by the next non-black start power node to be started; The recovery path charging time constraint of the non-black start power supply node is specifically: Where, Indicates the startup time of the mth non-black start power node to be restored, that is, the sth oblack,m The startup time of a non-black start power node, represents the startup time of the jth restored non-black start power node, i.e., the sth nblack,j The startup time of a non-black start power node, represents the charging time of the restoration path from the jth restored non-blackstart power node to the mth non-blackstart power node to be restored; The power constraints of the non-black start power supply node are specifically: Where, P G,k represents the active power output of the kth non-black start power node, represents the minimum active power allowed to be output by the i-th non-black start power node, represents the maximum active power allowed to be output by the kth non-black start power node; Q G,k is the reactive power generated by the kth non-black start power node, is the minimum reactive power allowed to be output by the kth non-black start power node, is the maximum reactive power allowed to be output by the kth non-black start power node, n G is the number of non-black start power supply nodes.

5. The power supply method for optimizing black start power distribution according to claim 4, characterized in that: The growth pattern of coyote individuals in the improved coyote optimization algorithm described in step 3 is as follows: Step 3.1: Initialize the coyote population: Set the coyote population size to N and the coyote individual dimension to D. The social state of the coyotes represents the decision variable of the optimization problem. The social state of the cth coyote individual in the pth group during the tth iteration is expressed as: Where t represents the current iteration number; represents the black start power supply installation location of the c-th individual solution vector in the coyote population during the t-th iteration, represents the black start power supply installed capacity of the c-th individual solution vector in the coyote population during the t-th iteration; Divide the coyotes into Np groups, and the number of coyotes in each group is Nc, then N = N p ×N c , initialize the coyote; Where lb j represents the lower limit of the j-th dimension of the coyote's social state, ub j represents the upper limit of the j-th dimension of the coyote's social state, r j is a random number in [0,1]; Calculate and evaluate the fitness of individual coyotes: in, represents the adaptability of the c-th coyote in the p-th group at the t-th iteration, represents the social state of the cth coyote individual in the pth group at the tth iteration, and f(*) represents the objective function with the shortest black start recovery time of the non-black start power supply node; Step 3.2: In-group coyote growth: Find the optimal individual in each coyote group: Among them, alpha p,t represents the leader of the p-th group at the t-th iteration, represents the social state of the c-th coyote in group p at the t-th iteration, Indicates the variable value when the objective function takes the minimum value; The coyotes in the group are arranged according to their fitness to obtain the Op,t sequence. The median of this sequence is used to evaluate the group culture trend of the coyote group, representing the average social situation of the coyote group: in, represents the cultural trend of the coyote group at the tth iteration of group p, Indicates that when N c When is an odd number, the median of the j-th dimension of all coyotes in the p-th group at the t-th iteration, Indicates that when N c When is an even number, it is the median of the jth dimension of all coyotes in group p at iteration t; Use the best individuals and cultural trends within the group to guide the growth of individual coyotes within the group: Where alpha p,t represents the optimal individual coyote in group p at the tth iteration, cult p,t represents the cultural trend within group p at iteration t, and are two coyotes randomly selected from group p at the tth iteration. r1 and r2 represent the first and second growth control parameters, respectively, and are random numbers in [0,1]. δ1 represents the optimal individual influence factor within the group, and δ2 represents the cultural trend influence factor within the group. represents the social state of the cth coyote in group p at the tth iteration after the coyote grows up; The social adaptability of coyotes before and after growth is evaluated based on the fitness function, and the social status of coyotes with better adaptability is retained; in, represents the adaptability of the c-th coyote in group p after it grows up at the t-th iteration. represents the social state of the c-th coyote in group p at the t-th iteration after the coyote grows up, and soc represents the social state of the c-th coyote in group p with better adaptability at the t-th iteration; The global optimal coyote influence factor δ3 is added to the growth model of the coyote optimization algorithm. This factor can guide the individuals in the group to grow in the direction of the global optimal coyote so far, thereby improving the global optimization ability and convergence speed of the algorithm. The improved growth method of the coyote group is: Where alpha p,t represents the optimal individual coyote in group p at the tth iteration, cult p,t represents the cultural trend of group p at iteration t, galpha p,t represents the globally optimal coyote of group p at iteration t, and are three coyotes randomly selected from group p at the tth iteration, r1, r2, and r3 represent the first, second, and third growth control parameters, respectively, and are all random numbers in [0,1]; Step 3.3: Coyote birth and death: The birth of a new coyote is the result of the union of two randomly selected parent coyotes within the group, plus the combined effects of environmental factors; Where, represents the newly born coyotes in the jth dimension of the pth group at the tth iteration, and They represent the two parent coyotes randomly selected from the jth dimension in the pth group at the tth iteration, rnd j is a random number in [0,1], R j is the randomly generated variation value within the j-th dimension of the coyote social state; P s =1 / D P a =(1-P s ) / 2 Where, P s is the inheritance probability, P a represents the probability of mutation, and D represents the individual dimension of the coyote; The Mantegna algorithm is used to generate random step sizes that obey the Levy distribution, specifically: Where s represents the random step size of the Levy distribution, γ represents the Levy parameter, μ represents the random number of the Gaussian distribution, and ν represents the random number of the Gaussian distribution. Among them, σ μ represents the variance of random number μ, σ v represents the variance of the random number ν, Γ represents the gamma function, which is the extension of the factorial function on real and complex numbers, and NO represents the normal distribution; The set S of flight steps in each dimension of the social state of coyote individual c c =[s1,s2,…,s D ], the coyote performs Lévy flight with a certain flight probability during the iteration process: Among them, rnd represents a random number, represents the social state of the cth coyote in group p at iteration t, P l represents the probability of Levy flight, s d Represents the d-th subvector in the flight step set, d∈[1,D]; Increases Levi's chance of flying, specifically: Among them, P l represents the probability of Levy flight, P min represents the minimum value of the Levy flight probability, P max represents the maximum value of Levy flight probability, n represents the number of iterations, n max Represents the maximum number of iterations; Adopting the elite flight exemption strategy, that is, the current global optimal coyote individual does not participate in the Levy flight and retains its corresponding social state; Step 3.4: Coyote Migration: Coyote populations do not remain static after being randomly grouped. Migration sometimes occurs between coyote groups. This ensures diversity within the coyote groups while also allowing for information exchange between different groups. Where, P e represents the probability that a coyote is driven away by a group and accepted by another group, N c represents the size of coyotes in the group, and ε represents the probability coefficient of coyote migration.

6. The power supply method for optimizing black start power distribution according to claim 5, characterized in that: The adaptive Levy flight described in step 3 is as follows: The Mantegna algorithm is used to generate random step sizes that obey the Levy distribution, specifically: Among them, γ represents the Levy parameter, μ represents the random number obeying the Gaussian distribution, and ν represents the random number obeying the Gaussian distribution; Among them, σ μ represents the variance of random number μ, σ v represents the variance of the random number ν, Γ represents the gamma function, which is the extension of the factorial function on real and complex numbers, and NO represents the normal distribution; The set S of flight steps in each dimension of the social state of coyote individual c c =[s1,s2,…,s D ], the coyote performs Lévy flight with a certain flight probability during the iteration process: in, represents the social state of the cth coyote in group p at iteration t, P l represents the probability of Levy flight, s d Represents the d-th subvector in the flight step set, d∈[1,D]; Increases Levi's chance of flying, specifically: Among them, P l represents the probability of Levy flight, P min represents the minimum value of the Levy flight probability, P max represents the maximum value of Levy flight probability, n represents the number of iterations, n max Represents the maximum number of iterations; An elite flight exemption strategy is adopted, that is, the current global optimal coyote individual does not participate in Levy flight and retains its corresponding social state.

7. The power supply method for optimizing black start power distribution according to claim 6, characterized in that: The chaos optimization perturbation mechanism described in step 4 is as follows: Set the chaotic mutation probability p h , randomly generate a real number in [0,1], if it is less than p h Then perform chaotic perturbation, otherwise directly enter the next iteration; Map each dimension of the global optimal coyote individual at time t to the chaotic space, specifically: Where, is the j-th dimension chaos value of the optimal coyote in the t-th iteration, is the j-th dimension value of the optimal coyote in the t-th iteration; and are the upper and lower limits of the j-th dimension of the coyote, respectively; Perform N-1 Logistic chaotic searches to obtain a chaotic sequence in, is the j-th chaotic value of the n-th chaotic variable of the global optimal coyote in the t-th iteration, is the j-th chaotic value of the n+1-th chaotic variable of the global optimal coyote in the t-th iteration, is the j-th dimension value of the global optimal coyote in the t-th iteration, represents the upper limit of the j-th dimension of the coyote, represents the lower bound of the j-th dimension of the coyote, n∈[1,N]; Map the N chaotic variables in the chaotic sequence to the solution space of the optimization problem according to the following formula to generate N new coyote individuals; in, represents the j-th dimension value of the n-th chaotic variable of the global optimal coyote at the t-th iteration, represents the upper limit of the j-th dimension of the coyote, represents the lower bound of the j-th dimension of the coyote, z n,j represents the j-th chaotic value of the n-th chaotic variable of the global optimal coyote; The adaptability of the new coyote individuals is evaluated according to the fitness function, and the coyote individuals with the best adaptability are retained, which is recorded as by The coyote replaces the coyote with the worst adaptability at time t.

8. A computer-readable medium, characterized in that It stores a computer program executed by an electronic device, and when the computer program runs on the electronic device, the electronic device executes the steps of the method according to any one of claims 1 to 7.

Images