Fast reconfiguration method for multi-load independent power distribution system based on pre-selection mechanism

Abstract

The application discloses a kind of multi-load independent power distribution system fault fast reconfiguration method based on preselection mechanism, belong to the field of fault reconfiguration, the present application aims at solving in complex power distribution system, due to the difference of power supply priority, the problem that anti-hitting ability demand leads to that fault reconfiguration algorithm is difficult to obtain high-quality reconfiguration scheme, the present application includes the following steps: obtaining the actual power grid structure of fault multi-load independent power distribution system and carrying out network structure coding, to be used as fault reconfiguration target.According to demand, construct initial fault reconfiguration mathematical model, define objective function and constraint condition, obtain heuristic model after integration.Input network structure coding into the model, optimization is carried out using preselection mechanism, and initial solution set is obtained.Optimal reconfiguration scheme solution set is obtained by multiple rounds of iteration optimization, and is further screened in combination with actual demand, and finally forms fault reconfiguration scheme.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fault reconstruction, and in particular to a multi-load independent power distribution system fault rapid reconstruction method based on a pre-selection mechanism. BACKGROUND

[0002] With the emergence of distributed energy, the power system is accelerating evolution, leading to more complex power distribution systems and more variable operating environments, and higher requirements for system adaptability. Multi-load independent power distribution system is a significant development in this regard, which includes a relatively small medium-sized power grid and carries a large number of loads. The system operates independently of large land-based regional power distribution networks and is widely used in various military and civilian facilities, including dedicated vehicles, ships and refuge systems, etc. In actual operation, due to wear and tear or improper operation, equipment or even systems may fail. The role of fault reconstruction is reflected in the self-repair operation of the system power distribution network after the power distribution system is damaged, the maximum rapid recovery of power supply to important loads, or the provision of corresponding auxiliary decision recommendations in other fault states, to ensure the safe and reliable operation of the power system.

[0003] Fault reconstruction of power distribution systems is usually converted into a single-objective or multi-objective optimization problem, so the current method for solving the reconstruction problem mainly relies on heuristic algorithms. These algorithms identify the optimal solution from random candidate solutions in a specific space, and the variation is due to the difference in the formula of the search method. The fault reconstruction of the power distribution system is abstracted as an objective function, and the heuristic algorithm is used to solve the objective function, and the optimal solution obtained is the reconstruction scheme of the system. Due to the efficiency and accuracy of heuristic algorithms, it is usually used for fault reconstruction of land-based regional power distribution systems or other small and medium-sized power distribution systems.

[0004] The current application of heuristic algorithms does not take into account the adverse effects of the characteristics of multi-load independent power distribution systems on reconstruction under fault conditions and the corresponding solutions. Compared with land-based regional power distribution systems, multi-load independent power distribution systems also have the following problems in fault reconstruction:

[0005] (1) The connection between power sources and loads in the system is closer, the number of loads is larger, the power supply priority of different loads is different, part of the important load is supplied by the main line and the standby line, and the network structure is more complex. When the power distribution network is abstracted as a mathematical problem, it is difficult to effectively reflect the original network structure characteristics of the power distribution system.

[0006] (2) Since the system bears a large number of loads with different power supply priorities, and the application scenario itself requires the system to have a certain anti-impact capability to cope with subsequent fault impact. Therefore, after taking these considerations into account, the objective function of the reconstruction becomes complex, and the current fault reconstruction algorithm is difficult to obtain a high-quality reconstruction scheme.

[0007] (3) Heuristic algorithm requires high quality of initial solution population, and the quality of initial solution population influences the solving speed and accuracy of the algorithm, and the influence is more obvious for complex solving problems such as fault reconstruction of a multi-load independent power distribution system. SUMMARY

[0008] In view of the deficiencies of the prior art, the present application provides a multi-load independent power distribution system fault fast reconstruction method based on a pre-selection mechanism, comprising the following steps:

[0009] Step 1: obtaining the actual power grid structure of the fault multi-load independent power distribution system;

[0010] Step 2: network structure coding is performed on the actual power grid structure of the fault multi-load independent power distribution system, and the network structure coding is the target of fault reconstruction;

[0011] Step 3: constructing an initial fault reconstruction mathematical model according to the requirements of fault reconstruction, and defining a target function and constructing a constraint condition;

[0012] The target function includes: relative load , switch operation cost and line capacity margin ;

[0013] The constraint condition includes: radial operation constraint and capacity constraint;

[0014] Determine the relative load (Relational Load) :

[0015] ;

[0016] In the formula, I1, I2 and I3 respectively represent the rated current of the primary, secondary and tertiary load; , , respectively represent the number of primary, secondary and tertiary loads; represents the load connected to the power grid or not connected to the power grid, and "1" represents access to the power grid and "0" represents not access to the power grid; respectively represent the number of primary, secondary and tertiary loads; respectively represent the maximum load value in the secondary and tertiary load, respectively represent the minimum load value in the primary and secondary load;

[0017] Determine the switch operation cost :

[0018] ;

[0019] In the formula, and respectively represent the number of primary, secondary and tertiary loads. are the cost weights of the switching operations required by the automatic switch and the manual switch, respectively; , and are the cost weights of the opening and closing operations required by the automatic switch, the manual switch and the other type switch, respectively; , , , and represent the number of each type of switch operation in the process of fault reconfiguration of the multi-level intrusion detection system (MLIDS);

[0020] Determination of line capacity margin :

[0021] ;

[0022] wherein, represents the actual load of the first important branch; represents the rated load of the first important branch; represents the total number of important branches in the power distribution system; represents the first three minimum elements; represents the average value;

[0023] Determination of radial operation constraints:

[0024] ;

[0025] wherein, represents the set of switching switches; and represent the on-off state of the primary and standby switches of the same load, wherein "0" represents that the switch is off, and "1" represents that the switch is on;

[0026] Determination of capacity constraints:

[0027] ;

[0028] wherein, represents the connection state of the load and the branch or the connection switching state of the branch and the distribution panel ; is the power consumption of the load or the branch, is the maximum capacity of the branch or the generator, is the number of loads contained in the requested branch ;

[0029] wherein, the and maximizing, minimizing;

[0030] Step 4: integrating the objective function and the constraint condition into the initial fault reconstruction mathematical model to obtain a final fault reconstruction mathematical model;

[0031] Step 5: inputting the network structure code into the final fault reconstruction mathematical model and optimizing the same by using a pre-selection mechanism to obtain an initial solution set ;

[0032] Step 6: based on the initial solution set , performing multi-round iteration optimization by using the fault reconstruction mathematical model to obtain an optimal reconstruction scheme solution set;

[0033] Step 7: after the schemes in the optimal reconstruction scheme set are further screened according to actual requirements, a final fault reconstruction scheme is formed.

[0034] As a further improvement of the present application, the pre-selection mechanism is used in step 5 to optimize the final fault reconstruction mathematical model to obtain an initial solution set , which includes an initialization stage, an individual generation stage and a selection stage.

[0035] As a further improvement of the present application, the initial population of the multi-objective optimization in the fault reconstruction mathematical model is set to , the population size is set to , and a large-scale population is constructed by using the pre-selection mechanism .

[0036] As a further improvement of the present application, the initial solution set obtained in step 5 specifically includes the following steps:

[0037] Step 5.1: selecting two high-priority objective functions as competition items for determining which individuals are suitable for being selected into the large-scale population and the initial population ;

[0038] Step 5.2: generating a plurality of random individuals ;

[0039] Step 5.3: selecting one of the competition items to test the random individuals , and the optimal individual is selected by the competition and added to the large-scale population ;

[0040] Step 5.4: checking whether the large-scale population requirement is met, if yes, entering step 5.5; if not, repeating step 5.3 until the number of selected individuals meets the large-scale population requirement.​ Request;

[0041] Step 5.5: Obtain large-scale population After that, test all individuals in the large-scale population using another competition project, and sort them according to their performance;

[0042] Since all individuals in the large-scale population are winners of the last competition project, they have shown good performance in the corresponding objective function, so the individuals are directly selected according to their performance in the second project;

[0043] Step 5.6: Select the top performing individuals to form the initial population of algorithm iteration , that is, the initial solution set .

[0044] As a further improvement of the application, the step 6 is specifically: based on the initial solution set , the meta-heuristic optimization algorithm is used to iteratively solve the fault reconstruction mathematical model to obtain the optimal reconstruction scheme solution set.

[0045] The beneficial effects of the application are:

[0046] The network structure matrix that more accurately reflects the network structure characteristics of the multi-load independent power distribution system is established, the connection between the reconstruction algorithm and the structure is strengthened, and the solving process is more in line with the characteristics of the system itself. The pre-selection mechanism effectively improves the quality of the solution set in the initial stage of the reconstruction algorithm. From the test results, the reconstruction algorithm after introducing the pre-selection mechanism has a more stable performance in dealing with the fault reconstruction problem of the multi-load independent power distribution system. Even if the problem complexity is increased, an excellent reconstruction scheme can be obtained.

[0047] By deeply studying the characteristics of the multi-load independent power distribution system, the objective function and the constraint condition of the fault reconstruction are optimized. A multi-load independent power distribution system fault fast reconstruction method based on a pre-selection mechanism is proposed to obtain a reconstruction scheme. Simulation experiments show that the integration of the innovative mechanism significantly improves the convergence speed and optimization ability of the fault reconstruction method, thereby achieving better reconstruction results in the fault reconstruction scenario of the multi-load independent power distribution system.

[0048] By combining the characteristics of the multi-load independent power distribution system and the network structure characteristics, a new encoding method that better reflects the circuit characteristics of the system is proposed, which facilitates the reconstruction algorithm to quickly find the optimal solution in the subsequent solving process. The pre-selection mechanism is added in the initial stage of the reconstruction algorithm to ensure that the reconstruction algorithm obtains an excellent reconstruction solution set in the initial stage, accelerates the convergence speed of the algorithm, and improves the performance of the algorithm in higher dimensions and more complex fault situations. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 Flow chart of the embodiment of the method for fast reconfiguration of multi-load independent power distribution system based on pre-selection mechanism;

[0050] Figure 2 Structure design diagram of the multi-load independent power distribution system of a certain large destroyer in the embodiment;

[0051] Figure 3 MLIDS schematic diagram of the multi-load independent power distribution system of a certain large destroyer in the embodiment;

[0052] Figure 4 Network structure diagram of the simple distribution system in the embodiment;

[0053] Figure 5 Initialization operation flow chart with pre-selection mechanism in the embodiment;

[0054] Figure 6 Distribution of Pareto front and solutions in the fault case in the embodiment;

[0055] Figure 7 Overall closeness of Pareto front in the fault case in the embodiment;

[0056] Figure 8 Influence of the maximum number of evolutions on the performance of each algorithm in the fault case in the embodiment. DETAILED DESCRIPTION

[0057] This embodiment is described by taking a certain ship ring power distribution network as an example;

[0058] The application provides a method for fast reconfiguration of multi-load independent power distribution system based on pre-selection mechanism, which can quickly and accurately obtain a fault reconfiguration scheme in combination with the characteristics of the system itself when the multi-load independent power distribution system fails. The fault reconfiguration scheme obtained by using the method is more suitable for the multi-load independent power distribution system, and when the requirements for the reconfiguration scheme are improved, a high-quality scheme result can also be obtained.

[0059] As shown in Figure 1 A method for fast reconfiguration of multi-load independent power distribution system based on pre-selection mechanism comprises the following steps:

[0060] Step 1: obtaining the actual power grid structure of the fault multi-load independent power distribution system;

[0061] As shown in Figure 2As shown in the figure, the multi-load independent power distribution system of the large destroyer adopts a ring structure design, which is composed of four generator sets (G1-G4) as the main power supply source of the ship, and is also equipped with a power supply network composed of twenty loads (L1-L20), which are divided into three different priority levels. The four generator sets are connected to each other through four tie cables (TL1-TL4). Eight area distribution panels (LCP1-LCP8) located below the main distribution panel are responsible for supplying power to low-priority loads. Some independent loads are directly connected to the main distribution panel through switches (L1, L6, L11 and L16). For relatively important primary and secondary loads (L1, L3, L5, L6, L8, L10, L11, L13, L15, L16, L18 and L20), the selection of their main and backup power supply paths is realized through automatic transfer switches (ATS) or manual transfer switches (MTS), thereby ensuring the continuous power supply of the loads.

[0062] For the convenience of analyzing the network structure, each branch and node in the multi-load independent power distribution system in Figure 1 is numbered, and the symbolic diagram of the above multi-load independent power distribution system is shown in Figure 3 .

[0063] Step 2: Network structure coding is performed on the actual power grid structure of the fault multi-load independent power distribution system, and the network structure coding is the target of fault reconstruction;

[0064] Network structure coding is an excellent means for processing the network structure of the power distribution system, which can convert the actual power grid structure into a digital object. These digital objects can accurately present the structure, composition and real-time state of the power grid. Although the common coding methods applied in the fault reconstruction of the power distribution system can effectively represent the network structure, they also have obvious limitations. Specifically, these methods mainly have three problems: first, the changes of the power grid structure are not fully reflected; second, infeasible solutions are not considered; third, the scalability is poor.

[0065] To solve the above problems and strengthen the relevance of coding and the actual MLIDS network structure, the present application proposes a new network structure coding scheme based on the branch correlation matrix method. The network structure coding scheme can effectively verify the network connectivity, radial operation constraints and capacity constraints. As shown in Figure 4 , a simple power distribution system network containing nodes, branches and loads is used to show the coding method proposed by the present application.

[0066] The branch correlation matrix method is a technology that uses matrices to describe the network structure of the power distribution system and the direct connection path between the load and the power supply. When the network structure becomes complex, it becomes difficult to clearly describe the system network structure with matrices. Therefore, when constructing the branch correlation matrix, it is necessary to identify the paths that can promote power transmission from the load perspective.Figure 4 The branch incidence matrix of the network structure can be represented as follows:

[0067] ;

[0068] The non-zero elements in the first row of the matrix represent the loads, the elements in the corresponding positions in each column represent the paths that can be connected to the power supply, the elements in the columns with zero values in the first row correspond to the alternative paths on the left side of the load, and the non-zero elements at the end of each column represent the direct connection branches of the power supply. By using a new encoding method to encode the MLIDS network topology in the matrix, a complete branch incidence matrix is finally obtained. Figure 2

[0069] ;

[0070] ;

[0071] The final stage of the optimization process, especially the reconstruction of the multi-load independent power distribution system, will generate a series of switch operations or load power supply states, forming a discrete data set. Therefore, when an evolutionary algorithm is used for ship power grid reconstruction, the solution form must be discretized.

[0072] The present application adopts a 0-1-2 encoding rule: the operating state of a standard load is encoded as 0-1, where "0" represents that the load stops supplying power, and "1" represents that the load starts supplying power; for important loads with a backup power supply path, the system uses a 0-1-2 encoding mode, where "0" represents that the load is stopped, "1" represents that the main path supplies power, and "2" represents that the backup path supplies power.

[0073] Step 3: Construct an initial fault reconstruction mathematical model according to the requirements of fault reconstruction, and define the objective function and construct the constraint conditions;

[0074] An initial fault reconstruction mathematical model of the multi-load independent power distribution system is constructed, and the objective functions and constraint conditions such as relative load, switch operation cost, line capacity margin, radial operation constraint, capacity limit, and line capacity upper limit are defined. By comprehensively considering the above objective functions and constraint conditions, a complete fault reconstruction mathematical model is constructed. This model can optimize the objective function on the basis of meeting all the constraint conditions, and then explore the optimal fault reconstruction scheme.

[0075] In view of the limitation of the load capacity of the generator in the multi-load independent power distribution system, and the existence of key lines in the system, when the system capacity is insufficient, unnecessary loads need to be determined as abandoned objects.

[0076] ​In the construction of the fault reconfiguration mathematical model, the capacity constraints and load priority of the system need to be considered comprehensively. The model takes the maximization of the total power of the necessary load recovered as the objective function, while meeting the output current limit of each generator and the capacity limit of the key line.

[0077] The core goal of the fault reconfiguration mathematical model is to accelerate the recovery process of the power distribution system while ensuring the continuity and reliability of the power supply to critical loads. To this end, the fault reconfiguration mathematical model sets the relative load, switch operating cost, and line capacity margin as the objective function, and is subject to three constraints: the power supply line network structure, system capacity, and upper limit of line capacity margin.

[0078] The objective function and constraint setting method are as follows:

[0079] Determine the relative load (Relational Load) :

[0080] ;

[0081] In the formula, , , represent the rated current of the first, second, and third load, respectively; represents the load connected to the power grid or not connected to the power grid, "1" represents access to the power grid, and "0" represents not access to the power grid; represent the number of first, second, and third load, respectively; represent the maximum load value in the second and third load, respectively, represent the minimum load value in the first and second load, respectively.

[0082] The relative load can accurately reflect the contribution of different priority loads to the total load. When performing reconfiguration operations, the load with a higher relative load value is preferentially selected for configuration.

[0083] Determine the switch operating cost :

[0084] ;

[0085] In the formula, and are the cost weights required for automatic switch and manual switch to perform switching operations; , and are the cost weights required for automatic switch, manual switch, and other types of switches to perform opening and closing operations; , , , and respectively represent the number of each type of switch operation in the process of multi-level intrusion detection system (MLIDS) fault reconfiguration. Meanwhile, in the reconfiguration process, the switch operation cost should be kept at the lowest possible value to minimize the power supply recovery time.

[0086] Determination of line capacity margin

[0087]

[0088] wherein, represents the actual load of the first important branch; represents the rated load of the first important branch; represents the total number of important branches in the distribution system; represents the first three minimum elements; represents the average value.

[0089] The line capacity margin is defined as the ratio of the remaining load to the rated value after the branch is loaded. If the minimum ratio branch still has a large capacity margin, it can be inferred that the distribution system under the reconfiguration scheme still has the ability to access the load and improve its ability to resist subsequent shocks. Therefore, it is necessary to ensure that the system has sufficient line capacity margin during the reconfiguration process.

[0090] Determination of radial operation constraints:

[0091]

[0092] wherein, represents the set of switching switches; and respectively represent the on-off state of the primary and standby switches of the same load, wherein "0" indicates that the switch is open, and "1" indicates that the switch is closed.

[0093] The radial operation constraint avoids the formation of loops and double simultaneous power supply between loads.

[0094] Determination of capacity constraints:

[0095]

[0096] wherein, represents the connection state of the load and the branch or the connection switching state of the branch and the distribution panel ; is the power consumption of the load or branch, is the maximum capacity of the branch or generator, is the number of loads contained in the requested branch .​​​​

[0097] Capacity constraints play a key role in avoiding overloading of corresponding branches and generators due to load transfer.

[0098] Step 4: Integrate the objective function and constraints into the initial fault reconfiguration mathematical model to obtain the final fault reconfiguration mathematical model;

[0099] The final mathematical model of fault reconfiguration configuration is as follows:

[0100] ;

[0101] Step 5: Input the network structure code into the final fault reconfiguration mathematical model, and optimize it using the pre-selection mechanism to obtain an initial solution set ;

[0102] In a multi-objective optimization problem, multiple sub-targets need to be optimized simultaneously. The optimal values of each objective function in the fault reconfiguration mathematical model often contradict each other and it is difficult to find a solution that can simultaneously satisfy the optimal values of all sub-target functions Therefore, the Pareto optimal solution is usually used to handle the inherent trade-off between multiple objectives in multi-objective optimization.

[0103] If the individuals p and q in the decision space satisfy the conditions in equation (7), it means that for any given set of objective functions, there is an individual p that satisfies p is not worse than q, and p is better than q in at least one objective. At this time, p dominates q, denoted as In the feasible solution set of a multi-objective optimization problem, if there is no such that , then is called the Pareto optimal solution (also known as non-dominated solution) of the problem. The set composed of the objective function vectors corresponding to the solutions in the optimal solution set is designated as the Pareto front.

[0104] ;

[0105] In the fault reconstruction process of MLIDS, each individual in the solution space represents the state of the switch quantity and is discretely distributed. This phenomenon causes the differences between different solutions to exhibit a hierarchical distribution characteristic, and this difference is more significant in the initialization stage of the algorithm. A weak initial population may adversely affect the subsequent performance of the algorithm.

[0106] To strengthen the initialization stage of the fault reconstruction model and accelerate its convergence speed, a pre-selection mechanism is proposed to help the model quickly obtain high-quality initial population. The pre-selection mechanism includes using a multi-objective optimization algorithm to preliminarily screen multiple solution sets in the model startup stage. Then, according to the actual requirements, those solutions that are more conducive to the subsequent evolution of the population are selected to form the initial population, thereby helping the model quickly obtain a better reconstruction scheme set. This mechanism helps to improve the efficiency and effectiveness of the algorithm and ensures that the optimal solution can be approached faster in the subsequent iteration process.

[0107] Specifically, assuming that the initial population of multi-objective optimization in the fault reconstruction mathematical model is , the population size is , a large-scale population is constructed by the pre-selection mechanism, which includes the initialization, individual generation, and selection stages, to prepare for subsequent competition and selection.

[0108] The initialization stage is the process of obtaining a larger population, which helps to search for the best individual more extensively, and specifically includes: selecting some objective functions for competition, generating a sufficient number of random individuals.

[0109] Individual generation stage:

[0110] Before selecting the initial population for algorithm iteration, a "preliminary screening" is needed to obtain the competition qualification: first, a group of individuals is randomly generated, a competition project is selected to evaluate these individuals, and the best individual is selected into the large-scale population ; repeat this operation until the number of selected individuals meets the requirements of the large-scale population.

[0111] Selection stage:

[0112] After obtaining the large-scale population , an additional competition program is used to evaluate all individuals; given that these individuals have performed well in the initial stage and have achieved significant results in related objective functions, they will be directly selected according to subsequent performance screening, and the top NP optimal individuals are selected to form the initial population for algorithm iteration. The selection of the competition project reflects the pre-selection standard, which usually depends on the characteristics of the underlying problem and the expected results.

[0113] The initialization operation using the pre-screening mechanism can generate a large number of high-quality initial individuals, enabling the algorithm to converge more efficiently to the current optimal solution.

[0114] where the initial solution set is obtained, specifically including the following methods:

[0115] Assuming that the initial population of multi-objective optimization in the fault reconstruction mathematical model is , the population size is , build large population through pre-selection mechanism , prepare for subsequent competition and selection;

[0116] Step 5.1: Select two high-priority objective functions as competition items to determine which individuals are suitable for selection into the large population and initial population ;

[0117] Step 5.2: Generate a number of random individuals ;

[0118] Step 5.3: Test the random individuals using one of the competition items, and the best individuals are selected through competition to join the large population ;

[0119] Step 5.4: Check if the large population requirements are met, if met, go to step 5.5; if not, repeat step 5.3 until the number of selected individuals meets the large population requirements;

[0120] Step 5.5: After obtaining the large population , test all individuals in the large population using the other competition item, and sort them according to their performance;

[0121] Since the individuals in the previous competition item have already shown good performance in the corresponding objective function, they are directly selected according to their performance in the second item;

[0122] Step 5.6: Select the top individuals with excellent performance to form the initial population for algorithm iteration, which is the initial solution set .

[0123] Step 6: Based on the initial solution set , use meta-heuristic optimization algorithms to iteratively solve the fault reconstruction mathematical model, and obtain the optimal reconstruction scheme solution set.

[0124] Based on the initial solution set , use meta-heuristic optimization algorithms to iteratively solve the fault reconstruction mathematical model. During the solving process, new candidate solutions are generated through selection, crossover, and mutation operations, and candidate solutions are evaluated according to the objective function. After meeting the convergence conditions, the final output is a reconstruction scheme solution set containing multiple optimal solutions.

[0125] Step 7: After further screening of the schemes in the optimal reconstruction scheme set according to actual requirements, the final fault reconstruction scheme is formed.

[0126] In practical applications, users will usually choose the most suitable solution according to their specific needs and preferences.

[0127] To verify the performance of the multi-load independent power distribution system fault fast reconstruction method based on the pre-selection mechanism proposed in the application, through typical fault scenarios in the multi-load independent power distribution system, the excellent optimization ability and convergence speed of the multi-load independent power distribution system fault fast reconstruction method based on the pre-selection mechanism in the fault reconstruction problem are demonstrated, and the stability of the method under various fault conditions is also demonstrated.

[0128] The multi-load independent power distribution system fault fast reconstruction method based on the pre-selection mechanism proposed in the application is compared with the hybrid optimization algorithm BHNM and the multi-objective optimizer NSGA-II-MC. The objective function and its constraint conditions are shown in formula (6). When the solution result violates the constraint condition, the objective function value will impose a penalty on it.

[0129] Before the fault occurs, the MLIDS is in the normal operation state shown in the figure, and the generator and each load remain normal independent operation, and the detailed information of the fault condition is as follows: Figure 2

[0130] Fault case: the fault causes the generator G2 to stop working, G1, G3 and G4 to work normally, and due to the fault, the lines 16 and 69 are disconnected, and the loads L4-L10, L14 and L15 lose power supply.

[0131] The parameters of each algorithm are configured as follows: the initial values of the fault reconstruction method of the application are 0.4, 0.65 and 0.4 respectively. The maximum evolution number of each algorithm is 80 times, and the fault case is independently run for 20 times.

[0132] Compared with single-objective optimization problems, the optimal solution set of multi-objective optimization problems is more massive. Existing algorithms often have difficulty in completely converging to the Pareto front. Therefore, the algorithm performance is evaluated from four aspects: the distribution characteristics of the final solution in the objective space, the overall proximity to the Pareto front, the solution excellence rate, and the influence of the maximum evolution number on the algorithm. Among them, the overall proximity to the Pareto front and the solution excellence rate are defined as follows:

[0133] ;

[0134] In the formula, the definition is the overall proximity to the Pareto front, is the objective value in the Pareto front, is the objective value of the final solution of the algorithm. ​The function is used to calculate the distance between two points. The smaller the overall closeness to the Pareto front, the closer the final solution is to the Pareto front. This indicates that the algorithm has improved in its ability to find optimal solutions.

[0135] Solution premium rate

[0136] ;

[0137] wherein represents The excellent rate of the final solution obtained by the algorithm; represents the total number of final solutions obtained by the algorithm; represents the number of final solutions converging to the Pareto front, represents the number of Pareto optimal solutions under the current fault scenario. The algorithm is used to measure the ability of the algorithm to identify optimal solutions, while reflects the diversity of the solution. The goal is to maximize the excellent rate of the solution, with a target value of 1, representing the optimal quality.

[0138] In the fault case, the Pareto optimal solution set contains a total of 14 non-dominated solutions, and the specific distribution of the Pareto front and the cumulative distribution of the final solution of each algorithm in the target space are shown in Figure 6 .

[0139] The case results in nearly half of the load being powered off, and the number of feasible solutions is also greatly reduced. However, the present application performs well at this stage and successfully identifies almost all of the Pareto optimal solutions. As shown in Figure 8 and listed in Table 1, the box plot comprehensively presents the overall closeness of the boundary of each algorithm in 20 running tests, as well as the excellent rate distribution of the solution.

[0140] Table 1 Excellent rating of solution (=14) ;

[0141] ;

[0142] As shown in Figure 7 , the vast majority of final solutions obtained by the present application can directly converge to the Pareto front. Combined with the data in Table 1, it can be seen that compared with the NSGA-II-MC algorithm, the BHNM algorithm exhibits better performance under more stringent fault conditions. It is worth noting that in the fault scenario, the excellent rate of the present application always maintains a significant advantage.

[0143] As shown in Figure 8As shown, the severity of the fault case leads to performance instability and dramatic fluctuations of BHNM and NSGA-II-MC at a high number of sensitization. In contrast, the multi-load independent power distribution system fault rapid reconstruction method based on the pre-selection mechanism proposed in the present application not only maintains the fastest convergence speed, but also operates stably. The test results of the fault case show that the present application performs well in the fault reconstruction of the MLIDS and can quickly and accurately provide the optimal reconstruction scheme for different fault levels.

[0144] The above-described embodiments only express several embodiments of the present application, which are described in detail and specifically, but should not be understood as a limitation on the scope of the patent. It should be noted that for ordinary skilled in the art, without departing from the concept of the present application, several modifications and improvements can be made, which are all within the scope of protection of the present application. Therefore, the scope of protection of the present application should be subject to the appended claims.

Claims

1. A method for rapid fault reconfiguration of a multi-load independent power distribution system based on a pre-selection mechanism, characterized in that, Includes the following steps: Step 1: Obtain the actual power grid structure of the faulty multi-load independent power distribution system; Step 2: Perform network structure coding on the actual power grid structure of the faulty multi-load independent power distribution system. The network structure coding is the target of fault reconstruction. Step 3: Construct an initial fault reconstruction mathematical model based on the requirements of fault reconstruction, and define the objective function and construct constraints; The objective function includes: relative load Switch operation cost and line capacity margin ; The constraints include: radial operation constraints and capacity constraints; Determine the relative load. : ； In the formula, , , These represent the rated currents of the primary, secondary, and tertiary loads, respectively. This indicates whether the load is connected to or not connected to the power grid; "1" represents connected to the power grid, and "0" represents not connected to the power grid. These represent the number of level 1, level 2, and level 3 loads, respectively. These represent the maximum load values ​​in the second and third level loads, respectively. These represent the minimum load values ​​in the first and second level loads, respectively; Determine the cost of switch operation : ； In the formula, and These are the cost weights required to perform switching operations for automatic and manual switches, respectively. , and The cost weights required to perform opening and closing operations for automatic switches, manual switches, and other types of switches are respectively assigned. , , , and These represent the number of times each type of switch operation is executed during the fault reconstruction process of the Multi-Level Intrusion Detection System (MLIDS). Determine line capacity margin : ； In the formula, This indicates the actual load of the first important branch; This indicates the rated load of the first major branch; This indicates the total number of important branches in the power distribution system; This represents the first three smallest elements; This represents the average value; Determine radial operating constraints: ； In the formula, Represents a set of toggle switches; and These represent the open / closed states of the primary and backup switches for the same load, where "0" indicates the switch is open and "1" indicates the switch is closed. Determine capacity constraints: ； In the formula, Indicates load With branches Connection status or branch With the distribution panel The connection switching status; It is the power consumption of the load or branch. It is the maximum capacity of the branch or generator. It is a request branch The number of loads included below; Among them, the and Find the maximum value. Find the minimum value; Step 4: Integrate the objective function and constraints into the initial fault reconstruction mathematical model to obtain the final fault reconstruction mathematical model; Step 5: Input the network structure encoding into the final fault reconstruction mathematical model, and optimize it using a pre-selection mechanism to obtain an initial solution set. ; Step 6: Based on the initial solution set The fault reconstruction mathematical model is solved iteratively to obtain the optimal reconstruction solution set; Step 7: After further screening based on actual needs, the solutions in the optimal refactoring solution set are used to form the final fault refactoring solution.

2. The method for rapid fault reconfiguration of a multi-load independent power distribution system based on a pre-selection mechanism according to claim 1, characterized in that, In step 5, a pre-selection mechanism is used to optimize the final fault reconstruction mathematical model to obtain an initial solution set. It includes: the initialization phase, the individual generation phase, and the selection phase.

3. The method for rapid fault reconfiguration of a multi-load independent power distribution system based on a pre-selection mechanism according to claim 2, characterized in that, In step 5, the initial solution set is obtained. First, the initial population for multi-objective optimization in the fault reconstruction mathematical model needs to be set as follows: Population size is Large-scale populations are constructed through a pre-selection mechanism. .

4. The method for rapid fault reconfiguration of a multi-load independent power distribution system based on a pre-selection mechanism according to claim 3, characterized in that, In step 5, the initial solution set is obtained. Specifically, it includes the following steps: Step 5.1: Select two high-priority objective functions as competition items to determine which individuals are suitable for inclusion in a large-scale population. and the initial population ; Step 5.2: Generate several random individuals ; Step 5.3: Select one of the competition items for a random individual. Tests are conducted, and the best individuals are selected through a competitive process to join a large-scale population. ; Step 5.4: Check if the requirements for a large-scale population are met. If they are, proceed to step 5.5; otherwise, repeat step 5.3 until the number of selected individuals meets the requirements for a large-scale population. Require; Step 5.5: Obtain a large-scale population Subsequently, another competition project was used to study large-scale populations. All individuals were tested and ranked according to their performance. because All individuals in the second project were winners of the previous competition and have demonstrated good performance in the corresponding objective function, so the selection is based directly on the performance of the individuals in the second project. Step 5.6: Select top performers The initial population for algorithm iteration consists of a number of individuals. That is, the initial solution set. .

5. The method for rapid fault reconfiguration of a multi-load independent power distribution system based on a pre-selection mechanism according to claim 1, characterized in that, Specifically, step 6 involves using a metaheuristic optimization algorithm to iteratively solve the fault reconstruction mathematical model to obtain the optimal reconstruction solution set.

Images