Ship power supply system fault reconstruction method based on genetic particle swarm hybrid optimization

Abstract

The invention relates to a ship power supply system fault reconstruction method based on genetic particle swarm hybrid optimization, and the method comprises the steps: building a simulation model based on a typical annular structure of a ship power supply network, and enabling the structure of the ship power supply network to cover a plurality of generators and different levels of power loads; normal and standby power supply lines are adopted to ensure the flexibility and redundancy of the system, priority ranking is performed on loads, and the continuity of key loads is ensured through dual-path power supply; according to the ship power supply network fault reconstruction method, branch faults or generator faults can be simulated, an improved self-adaptive hybrid genetic particle swarm algorithm is integrated into an established simulation model, a potential solution in a ship power supply network fault reconstruction problem serves as particles, an optimal post-fault configuration scheme is obtained, and network topology structure reconstruction after fault handling is completed. On the basis of being compatible with a traditional intelligent algorithm, the method has the advantages of being high in convergence speed, high in precision, high in stability and capable of achieving multi-node collaborative optimization. The method can be applied to research issues such as ship power system fault reconstruction monitoring, intelligent algorithm optimization, fault diagnosis and positioning and the like.

Description

Technical Field

[0001] The present invention relates to the field of ship power system detection, and in particular to a ship power supply system fault reconstruction method based on genetic particle swarm hybrid optimization. Background Art

[0002] The reconstruction strategy of ship power system often adopts a centralized structure to adapt to the reconstruction requirements of simple power supply network. With the increasing complexity of ship power grid, the continuous improvement of ship industry automation level and power electronics technology, higher requirements are placed on the reliability of ship power supply system. The existing methods can no longer meet the requirements of accuracy, speed, convergence and stability in fault reconstruction of complex integrated power system. Summary of the invention

[0003] Aiming at the problems of high randomness and poor convergence of fault reconstruction of ship power system and its intelligent algorithm, a fault reconstruction method of ship power system based on genetic particle swarm hybrid optimization was proposed.

[0004] The technical solution of the present invention is: a method for fault reconstruction of a ship power supply system based on genetic particle swarm hybrid optimization, which specifically includes the following steps:

[0005] 1) Establishment of simulation test model: A simulation model is established based on the typical ring structure of the ship power supply network. The ship power supply network structure covers multiple generators and different levels of power loads. Normal and backup power supply lines are used to ensure the flexibility and redundancy of the system, and the loads are prioritized. The key loads are powered by dual paths to ensure their continuity. The simulation model simulates branch failures or generator failures and supports testing of multiple failure scenarios.

[0006] 2) Improved adaptive hybrid genetic particle swarm algorithm: First, the Circle chaos map population is initialized; then, the crossover and mutation operations in the genetic algorithm are performed on the population generated by the particle swarm optimization algorithm; the adaptive nonlinear parameter update dynamically adjusts the inertia weight and acceleration coefficient parameters according to the current search state, and uses the adaptive nonlinear update strategy to adjust the search strategy at different optimization stages to improve the convergence speed;

[0007] 3) Fault reconstruction: The improved adaptive hybrid genetic particle swarm algorithm is integrated into the simulation model. The potential solutions to the ship power supply network fault reconstruction problem are taken as particles. The position of each particle in the search space corresponds to a specific set of switch states. The switch states define the topology of the power supply network, which constitutes a configuration scheme of the power supply network as a solution. The particle size represents the number of switches in the power supply network configuration scheme represented by each particle. The minimum load power loss and switching time loss constitute the objective function. The power supply network under fault conditions is reconstructed by the improved adaptive hybrid genetic particle swarm algorithm to obtain the optimal configuration scheme. The output minimizes the load power loss and the number of switch operations under fault conditions, thus completing the reconstruction of the network topology structure after the fault.

[0008] Further, the objective function in step 3) is:

[0009] f=w1×f1+w2×f1

[0010] Where: f represents the total loss; f1 and f2 are the load power loss and switching time loss respectively; w1 and w2 are the weight values ​​of the load power loss and switching time loss respectively;

[0011] In any case, according to the load priority order, the power supply to the first and second level loads shall be restored first, and the power loss of the loads shall be minimized. The load power loss is as follows:

[0012]

[0013] Where: α1, α2, α3 represent the power failure weights of level 1, level 2, and level 3 loads respectively; n1, n2, n3 represent the number of three types of loads; x i , x j , x k ∈{0, 1} represents the power supply status of the load, 0 represents power failure, and 1 represents power supply; L i , L j , L k Indicates the power value of the load;

[0014] Assuming that the switching time of all manual and automatic operations is consistent, the switching time loss is proportional to the number of switching transitions, and the switching time loss is expressed as:

[0015]

[0016] Where: h1 and h2 represent the number of primary and secondary critical loads and tertiary non-critical loads respectively; Hs1∈{0,1} represents the switching action of non-critical loads, 0 means disconnection and 1 means unchanged; Hs2∈{0,1} represents the switching action of critical loads, 0 means unchanged and 1 means switching to standby or disconnection.

[0017] Furthermore, in step 2), the Circle chaos mapping population is initialized as follows: first, initial individuals are generated and distributed through a series of systematic steps to ensure the uniformity of the population in the search space; second, the population size and dimension are defined, and a chaotic sequence with pseudo-randomness is generated according to the Circle chaos mapping formula; then, these sequences are mapped to the actual search space of the problem to form complete individuals, thereby constructing a diverse and evenly distributed initial population; finally, statistical analysis or visualization methods are used to test the uniformity of the population distribution to ensure its diversity and globality, provide a high-quality initial solution for the optimization algorithm, and help the algorithm converge to the global optimal solution.

[0018] Furthermore, the adaptive nonlinear parameter updating strategy in step 2) is as follows: by real-time analysis of the search state during the optimization process, the key parameters of the inertia weight and the acceleration coefficient are dynamically adjusted to achieve adaptive control of the algorithm; the parameters are flexibly adjusted using nonlinear functions, and the search strategy is adaptively switched at different optimization stages to improve the breadth of the global search and the convergence speed of the local search.

[0019] The beneficial effects of the present invention are as follows: the ship power supply system fault reconstruction method based on genetic particle swarm hybrid optimization can perform fault reconstruction performance test based on the ship ring power supply network simulation model, and on the basis of being compatible with traditional intelligent algorithms, it has the characteristics of fast convergence speed, high accuracy, strong stability and multi-node collaborative optimization capabilities. It can be applied to research topics such as ship power system fault reconstruction monitoring, intelligent algorithm optimization, fault diagnosis and positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flow chart of the improved adaptive hybrid genetic particle swarm algorithm in the method of the present invention;

[0021] Figure 2 A schematic diagram of mixed genetic crossover variation in the method of the present invention;

[0022] Figure 3 This is a schematic diagram of a faulty ring power supply network according to an embodiment of the present invention;

[0023] Figure 4A An optimal loss iteration graph for implementing the method of the present invention;

[0024] Figure 4B This is the optimal switch coding diagram after the method of the present invention is implemented. DETAILED DESCRIPTION

[0025] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0026] The present invention proposes a method for fault reconstruction of a ship power supply system based on genetic particle swarm hybrid optimization. A simulation model is established based on the typical ring structure of a ship power supply network. The ship power supply network structure covers multiple generators and different levels of power loads; normal and backup power supply lines are used to ensure the flexibility and redundancy of the system, and the loads are prioritized. The key loads are powered by dual paths to ensure their continuity; it is able to simulate branch failures or generator failures, integrate an improved adaptive hybrid genetic particle swarm algorithm into the established simulation model, test the adaptability of the algorithm under different failure modes, and achieve a highly simulated test environment through parameter setting and adjustment. When used, potential solutions to the ship power supply network fault reconstruction problem are used as particles to obtain the optimal configuration solution, completing the reconstruction of the network topology structure after the failure. The simulation model embedded with the improved adaptive hybrid genetic particle swarm algorithm also defines performance evaluation indicators including convergence speed, target loss and algorithm stability, and supports the simulation of multiple and multi-point fault scenarios to verify the effectiveness and robustness of the algorithm.

[0027] Fault reconstruction refers to the method of disconnecting and isolating the faulty lines after one or several lines fail, and using contact lines to reconstruct the network to form a new network topology.

[0028] The "particles" in the improved adaptive hybrid genetic particle swarm algorithm represent potential solutions to the ship power supply network fault reconstruction problem. The position of each particle in the search space corresponds to a set of specific switch states, which define the topology of the power supply network and thus constitute a possible configuration scheme of the power supply network.

[0029] The definition of particle size is related to the specific dimension of the problem. In the present invention, the particle size represents the number of switches in the power supply network configuration scheme represented by each particle. The particle dimension is defined as 20, that is, each particle contains 20 switch states, corresponding to the power supply path configuration of 20 loads in the power supply network. Each switch state can be 0 (disconnected) or 1 (connected), indicating whether the power supply path of the corresponding load is activated. The particle size is related to the number of loads to be considered in the power supply network. The dimension of each particle is the dimension of the problem space, reflecting the configuration status of the loads and paths in the power supply network.

[0030] The improved adaptive hybrid genetic particle swarm algorithm mainly consists of three parts: Circle chaos map population initialization, hybrid genetic crossover mutation operation, and adaptive nonlinear parameter update. The process is as follows: Figure 1 shown.

[0031] The algorithm is designed as follows:

[0032] 1. Circle chaos mapping population initialization is used to generate and distribute the initial individuals of the algorithm population, ensure the uniformity of the population in the search space, avoid local aggregation of the population, improve the global search capability, and provide a diversified and uniformly distributed initial solution. The particle swarm is obtained by initializing the population through Circle chaos mapping, providing a high-quality initial search space, which is conducive to improving the convergence accuracy of the algorithm.

[0033] Circle chaotic mapping population initialization: First, generate and distribute initial individuals through a series of systematic steps to ensure the uniformity of the population in the search space; second, define the population size and dimension, and generate pseudo-random chaotic sequences according to the Circle chaotic mapping formula; then, map these sequences to the actual search space of the problem to form complete individuals, and then construct a diverse and evenly distributed initial population; finally, use statistical analysis or visualization methods to test the uniformity of the population distribution to ensure its diversity and globality, provide high-quality initial solutions for the optimization algorithm, and help the algorithm converge to the global optimal solution.

[0034]

[0035] Where: x i is the chaotic value of the i-th iteration; a and b are control parameters; mod(x,y) is the remainder function of x on y; the number of chaotic iterations is t c .

[0036] 2. Hybrid genetic crossover mutation is used to combine the advantages of particle swarm algorithm and genetic algorithm. The crossover operation retains excellent genes and improves the overall fitness of the algorithm; the mutation operation increases population diversity and prevents the algorithm from falling into local optimality. The combination of the two enhances local search efficiency and improves the convergence speed and optimization accuracy of the algorithm.

[0037] Hybrid genetic crossover mutation strategy: Combining the advantages of particle swarm optimization algorithm (PSO) and genetic algorithm (GA), such as Figure 2 , by performing crossover and mutation operations on the population generated by PSO, excellent genes are retained and population diversity is introduced. The crossover operation inherits the excellent characteristics of the parent generation and improves the efficiency of local search; the mutation operation prevents the algorithm from falling into the local optimum by randomly changing individual genes. This strategy optimizes individuals in multiple generations of iterations to improve the convergence speed and optimization accuracy of the algorithm.

[0038]

[0039] Where: x i (t+1) The new position of individual i after t+1 iterations, p iis the historical optimal position of individual i, g is the global optimal position, ω is the inertia weight of PSO, c1 and c2 are the learning factors of PSO, r1 and r2 are the random numbers of PSO, are two parent individuals, α is the weight of the crossover operation, β is the mutation intensity, and δ is the random perturbation.

[0040] 3. Adaptive nonlinear parameter update dynamically adjusts parameters such as inertia weight and acceleration coefficient according to the current search status. Adaptive nonlinear update strategy is used to adjust the search strategy at different optimization stages to improve convergence speed and maintain high accuracy and robustness.

[0041] Adaptive nonlinear parameter update strategy: By analyzing the search status in the optimization process in real time, dynamically adjusting key parameters such as inertia weight and acceleration coefficient, the algorithm can be adaptively controlled. Nonlinear functions are used to flexibly adjust parameters, and search strategies are adaptively switched at different optimization stages to improve the breadth of global search and the convergence speed of local search, achieving the unity of high precision and high robustness.

[0042]

[0043] Where: c imax 、c imin They represent the maximum and minimum values ​​of the acceleration factor c, t is the current number of iterations, and T is the total number of iterations.

[0044] Simulation test model design

[0045] 1) Load priority sorting: In the simulation model, the loads are prioritized and divided into three levels: Level 1 is important load, Level 2 is less important load, and Level 3 is non-important load. Level 1 and Level 2 loads are powered by dual paths to ensure that critical loads can be continuously powered. Load priority settings are shown in Table 1:

[0046] Table 1

[0047]

[0048] 2) Model infrastructure: Figure 3 ,The fault simulation model is based on the typical ring structure of the ship power supply network, ,where G1, G2, G3, and G4 are four diesel generators with equal power levels, ,and L1~L20 are power loads; at the same time, the power supply lines are divided into normal and ,standby. The solid line in the figure is the conventional power supply line, and the ,dotted line is the standby power supply line, which ensures flexibility and redundancy in ,the event of a fault.

[0049] 3) Fault simulation and parameter adjustment: Model parameters such as load current and line capacity are set according to the actual needs of the ship power supply system to achieve a highly simulated test environment. At the same time, the simulation model simulates branch faults or generator faults, supports testing of multiple fault scenarios, and verifies the effectiveness, robustness, and adaptability of the algorithm under fault conditions of different complexity.

[0050] 4) Simulation method steps: Use Figure 3 The algorithm is verified based on the ring network structure and the load data shown in Table 1. The simulation assumes that one branch of generators G1 and G2 fails (branch 7 and branch 35 fail, as shown in Figure 4). w0 is used as the initial weight value in the adaptive process, and M is the maximum capacity of the branch. The fault reconstruction program is written in Python for simulation. All simulation parameters are shown in Table 2 below.

[0051] Table 2

[0052]

[0053] 5) Algorithm integration and testing: The improved adaptive hybrid genetic particle swarm algorithm is integrated into the simulation model. The algorithm is used to reconstruct the power supply network under fault conditions and output the minimized load power loss and switch operation times under fault conditions. At the same time, the simulation model defines performance evaluation indicators, including convergence speed, target loss, and algorithm stability, to quantitatively evaluate the performance of the algorithm.

[0054] Among them: In any case, according to the load priority order, the power supply to the first-level and second-level loads shall be restored first, and the power loss of the load shall be minimized to the maximum extent. The load power loss is as follows:

[0055]

[0056] Where: f1 represents the objective function value, α1, α2, α3 represent the power failure weights of level 1, level 2, and level 3 loads respectively, n1, n2, n3 represent the number of three types of loads, x i , x j , x k ∈{0, 1} represents the power supply status of the load (0 means power failure, 1 means power supply), L i , L j , L k Indicates the power value of the load.

[0057] To simplify the analysis, it is assumed that the switching time of all manual and automatic operations is consistent, and the switching time loss is proportional to the number of switching conversions. The switching time loss can be expressed as:

[0058]

[0059] Where: f2 represents the switching time loss; h1 and h2 represent the number of primary and secondary critical loads and tertiary non-critical loads respectively; Hs1∈{0,1} represents the switching action of non-critical loads (0 means disconnection and 1 means unchanged); Hs2∈{0,1} represents the switching action of critical loads (0 means unchanged and 1 means switching to standby or disconnection).

[0060] Therefore, for this multi-objective fault reconstruction optimization problem, the weighted method is used to transform it into a single-objective optimization problem, and the single-objective function is:

[0061] f=w1×f1+w2×f1

[0062] Where: f represents the total loss; w1 and w2 are the weight values ​​of load power loss and switching time loss respectively.

[0063] Based on the improved adaptive hybrid genetic particle swarm algorithm in the ship ring power supply network simulation model, it converges to the optimal value and tends to be stable after the third iteration. Figure 4A ; After the algorithm is iteratively stabilized, it can achieve the optimal switch encoding, such as Figure 4B After 30 iterations in the simulation experiment, the switch code is 102111021111111111111, indicating that the optimal strategy is to cut off the power supply to loads 2 and 7, and switch 3 and 8 from the original power supply path to the backup power supply path, thus completing the reconstruction of the network topology structure after the fault.

[0064] In addition, while ensuring the convergence accuracy remains unchanged, the average target loss of the algorithm is 400, the proportion of optimal results is 100%, the average number of iterations required for convergence is 5.60, and the variance of the number of iterations required for convergence is 1.88, which has higher convergence speed and convergence stability.

[0065] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for fault reconstruction of ship power supply system based on genetic particle swarm hybrid optimization, characterized in that: The specific steps include: 1) Establishment of simulation test model: A simulation model is established based on the typical ring structure of the ship power supply network. The ship power supply network structure covers multiple generators and different levels of power loads. Normal and backup power supply lines are used to ensure the flexibility and redundancy of the system, and the loads are prioritized. The key loads are powered by dual paths to ensure their continuity. The simulation model simulates branch failures or generator failures and supports testing of multiple failure scenarios. 2) Improved adaptive hybrid genetic particle swarm algorithm: First, the Circle chaos map population is initialized; then, the crossover and mutation operations in the genetic algorithm are performed on the population generated by the particle swarm optimization algorithm; the adaptive nonlinear parameter update dynamically adjusts the inertia weight and acceleration coefficient parameters according to the current search state, and uses the adaptive nonlinear update strategy to adjust the search strategy at different optimization stages to improve the convergence speed; 3) Fault reconstruction: The improved adaptive hybrid genetic particle swarm algorithm is integrated into the simulation model. The potential solutions to the ship power supply network fault reconstruction problem are taken as particles. The position of each particle in the search space corresponds to a specific set of switch states. The switch states define the topology of the power supply network, which constitutes a configuration scheme of the power supply network as a solution. The particle size represents the number of switches in the power supply network configuration scheme represented by each particle. The minimum load power loss and switching time loss constitute the objective function. The power supply network under fault conditions is reconstructed by the improved adaptive hybrid genetic particle swarm algorithm to obtain the optimal configuration scheme. The output minimizes the load power loss and the number of switch operations under fault conditions, thus completing the reconstruction of the network topology structure after the fault.

2. According to claim 1, the method for fault reconstruction of a ship power supply system based on genetic particle swarm hybrid optimization is characterized in that: The objective function in step 3) is: f=w1×f1+w2×f1 Where: f represents the total loss; f1 and f2 are the load power loss and switching time loss respectively; w1 and w2 are the weight values ​​of the load power loss and switching time loss respectively; In any case, according to the load priority order, the power supply to the first and second level loads shall be restored first, and the power loss of the loads shall be minimized. The load power loss is as follows: Where: α1, α2, α3 represent the power failure weights of level 1, level 2, and level 3 loads respectively; n1, n2, n3 represent the number of three types of loads; x i , x j , x k ∈{0, 1} represents the power supply status of the load, 0 represents power failure, and 1 represents power supply; L i , L j , L k Indicates the power value of the load; Assuming that the switching time of all manual and automatic operations is consistent, the switching time loss is proportional to the number of switching transitions, and the switching time loss is expressed as: Where: h1 and h2 represent the number of primary and secondary critical loads and tertiary non-critical loads respectively; Hs1∈{0,1} represents the switching action of non-critical loads, 0 means disconnection and 1 means unchanged; Hs2∈{0,1} represents the switching action of critical loads, 0 means unchanged and 1 means switching to standby or disconnection.

3. According to claim 1, the method for fault reconstruction of a ship power supply system based on genetic particle swarm hybrid optimization is characterized in that: The Circle chaos mapping population initialization in step 2) is as follows: first, initial individuals are generated and distributed through a series of systematic steps to ensure the uniformity of the population in the search space; second, the population size and dimension are defined, and a chaotic sequence with pseudo-randomness is generated according to the Circle chaos mapping formula; then, these sequences are mapped to the actual search space of the problem to form complete individuals, thereby constructing a diverse and uniformly distributed initial population; finally, statistical analysis or visualization methods are used to test the uniformity of the population distribution to ensure its diversity and globality, provide a high-quality initial solution for the optimization algorithm, and help the algorithm converge to the global optimal solution.

4. According to claim 1, the method for fault reconstruction of a ship power supply system based on genetic particle swarm hybrid optimization is characterized in that: The adaptive nonlinear parameter updating strategy in step 2) is as follows: by real-time analysis of the search state during the optimization process, the key parameters of the inertia weight and the acceleration coefficient are dynamically adjusted to achieve adaptive control of the algorithm; the parameters are flexibly adjusted using nonlinear functions, and the search strategy is adaptively switched at different optimization stages to improve the breadth of the global search and the convergence speed of the local search.

Images