Improved Particle Swarm Algorithm And Its Application In The Reactive Power Optimization Of Power System

The reactive power optimization of power system is a typical nonlinear optimization problem, which has many characteristics, such as constraints, multivariable and discreteness. When using the current optimization methods, there still have lots of problems, for example, low precision, long optimal time and difficult in finding a global optimal solution. So it is necessary for us to explore new methods or to improve strategies.The particle swarm optimization, a representative swarm intelligence algorithm, is a powerful and useful tool in solving the problem of reactive power optimization. But in terms of theoretical analysis and application research, the particle swarm optimization is a relatively new optimization technique, which is still in its primary stage. There are many problems worth studying, such as how to improve the algorithm’s ability to jump out of local optimal solution, how to improve accuracy and speed in solving high dimensional complex multimodal problems.Firstly, in view of the disadvantage of falling into local optimal solution, this thesis puts forward two improved strategies:the elite comprehensive learning strategy which including elite weight and the championship neighborhood topology strategy. Then the author combines them with the standard particle swarm algorithm, which forms the improved particle swarm optimization. This thesis uses the improved algorithm to solve the reactive power optimization problem in power system.Secondly, this thesis does a further research in the model and solving method of large-scale power networks. On the basis of the improved particle swarm optimization, this thesis puts forward a distributed cooperative particle swarm optimization, trying to solve the reactive power optimization of large-scale and complex power networks. On the basis of the grid partition, this method can solve the reactive power optimization of large-scale and complex power networks with a distributed and collaborative way. Compared with the traditional centralized optimization calculation, this method can not only reduce the complexity of the problem, speed up the calculation but also can deal with coordination problems of different regulating equipment for reactive power. It is good for the implementation of comprehensive reactive voltage control of the power system.Finally, this thesis applies the theoretical results to the wind power project in one city. Based on the investigation of the project, this thesis designs a set of comprehensive reactive voltage control system, and uses the distributed cooperative particle swarm optimization algorithm to complete the reactive power optimization. By comparing, we come to the conclusion that the comprehensive reactive voltage control system based on the distributed cooperative swarm optimization algorithm is better than the separated control system and the centralized calculation system...