A Modified Particle Swarm Optimization Algorithm And Its Application In Reactive Power Optimization

In the rapid development of science and technology in these years, artificial intelligence (AI) has become one of the frontier science that attracting researchers. Along with the development of technology and practice of theory research, artificial intelligence is no longer an ideal concept of fuzzy, but widely used in the human society. These progress is based on the continuously expand of human wisdom. The key of AI is the artificial intelligence algorithm, which is concerned top priority by the research scholars. Comparing to the traditional method, artificial intelligence is proved to have advantages in dealing with optimization problems.This thesis is focusing on the PSO algorithm, and the basic structure and function would be discussed. Besides, algorithm model transformation is carried out to make it more flexible and improve the performance. To prove the performance of the modified PSO algorithm, this algorithm would be used to optimize the power flow calculation system. Analysis of the original PSO algorithm is talked in the first part. Disadvantages of the PSO algorithm and suggestions for improvement would be discussed. Then this thesis gives some suggestions to improve the algorithm performance. Tests would be carried out to validate the performance of the algorithm. Then, analysis of existing power flow calculation is carried out, as well as pointing out its advantages and disadvantages. The modified PSO algorithm is employed to optimizing the power flow calculation and the result shows this is practical and feasible. Also, this thesis analyzes the applicability of the modified PSO algorithm in optimal power flow calculation, and put forward the deficiency and prospect.The conclusion can be described as follows:according to the disadvantages of basic PSO algorithm, this thesis introduces PSO-VM which uses variable inertia weight to adjust exploring capability in different stages; mutation is also carried out to help algorithm escaping local extreme. This improved algorithm is tested by two practical node systems in power system and the result shows that PSO-VM has better performance and feasibility, and it is more comprehensive when applying in practical problems.