Shipboard Power System Fault Reconstruction Method Research Based On Gaussian Dynamic Particle Swarm Optimization Algorithm

With the expansion of Shipboard Power System (SPS), it has become more and more important to keep SPS running safely and stably. Naval ship needs a strong distribution network management system to satisfy different requirements under different working conditions and the load changes caused by incidents, and SPS reconstruction is one of the most important tasks. SPS reconstruction problems come down to be single objective/multi-objective combinatorial optimization problems, including discrete variables. Therefore, it is necessary to develop optimization algorithms according to the charaeteristics of modern SPSs. Particle swarm optimization (PSO) algorithm is a kind of novel intelligent optimization algorithm. It has been sueeessfully applied to some fields due to its simplicity and easy implementation. This dissertation conducts systematical research work on population topologies of Gaussian dynamic particle swarm optimization(GDPSO) algorithm, which can adapt to the different situations of SPS reconstruction problem. The major research work is shown as follows:The characteristics of SPS and main operation conditions are analyzed. The connectivity and topological structure of an typical 20 nodes SPS model are analyzed. On this basis, the mathematical model of SPS reconstruction optimization problem is established, and the general research steps is given. It is the basis and starting point of the follow-up research work.A reconstruction method of SPS is proposed based on GDPSO algorithm and static topology. First, associations between population topologies and performance of GDPSO algorithm are analyzed based on graph theory. To evaluate the performance of algorithm, a few set of topology statistic parameters (Average Degree, Standard Deviation of Degree Distribution, et al.) are choosen for comparing the performance parameters such as global and local search ability, success rate, and the numerical stability and so on. Second, a static topology called rand-5 is choosed for improving performance of GDPSO algorithm on the basis of the study, and the improved algorithm is applied for SPS reconstruction problems. Numerical simulation results demonstrate the effectiveness of the proposed method.For more complex SPS reconstruction multi-objective optimization problems, a new GDPSO algorithm with various structures based on dynamic topologies is presented, for searching solutions of different working conditions. To increase the population diversity, population topologies are dynamically changed by comparing the population entropy change and fitness value of objective function after each iteration. Early ring topology is used to keep global search capability, and then the topologies of k=5 keeps a good searching momentum, later all topology is adopted to conduct fine exploitation. Numerical simulations on SPS reconstruction with multi-objectives are run under two different situations of a typical example to demonstrate the effectiveness of this algorithm by comparing with other algorithms.Aiming at the complexity of the large-scale SPS reconstruction multi-objective optimization problems, an improved scale-free GDPSO algorithm is proposed as the SPS multi-objective reconstruction method. It uses a modified Barabasi-Albert (BA) model as a self-organizing construction mechanism. The swarm population is divided into two parts:rand-5 topology and scale-free topology. The rand-5 topology is used to keep good exploration and exploitation capability at the same time. And the scale-free network topology makes the swarm population to grow according to the modified BA model, which can increase the diversity of population. Numerical simulations are run with a 20 nodes model and an expanded 60 nodes model for fault reconstruction, by choosing two optimization targets:load power supply maximum and switch operation minimum. The effectiveness of the methodology is proved to solve high dimensional complex optimization problems.SPS reconstruction methods affect the quickness and accuracy of the strategy execution. Research work in this dissertation shows that the population topology structures greatly influence SPS reconstruction methods based on GDPSO algorithm. It can meet the requirements of SPS reconstruction according to the reconstruction targets to choose the appropriate static topologies or adaptive dynamic topologies.