Particle Swarm Optimization Based On TS Fuzzy Inference

Particle Swarm Optimization(PSO)is a kind of new Swarm Intelligence Optimization algorithm.PSO algorithm has many advantages,such as,it's distributed,collaborative,selforganizing and easy to be implemented.These characteristics make PSO algorithm can solve complex problems in absence of global information.It also opens a new way for typical and complex problems.Because the current study at the initial stage,there is still a large possibility fall into local optimum when solving high-dimensional and intricate problems.A reasonable balance between exploration and exploitation is the key to guarantee the global search ability,and it is a research focus during recent years in this field.This paper improved PSO algorithm from two aspects,firstly,we proposed a new Takagi-Sugeno inference engine based Adaptive Quantum-behaved Particle Swarm Optimization(AQPSO)algorithm.The algorithm can dynamically adjust its parameters and iteration method based on population diversity and the number of iterations.This can ensure the population search in a larger space.It also can reduce the probability of falling into local optima;secondly,in order to improve the performance of PSO algorithm,We raised DARPSO algorithms based on ARPSO algorithm which put forward by Riget.It is not simple to adopt linear gradient strategy simply,but dynamically to change inertia weight by that the state of particle is attractive or repulsive.At the same time,we design a new update way of particle position.Several test functions simulation and the Wilcoxon signed rank test show that AQPSO can obtain better results when the values of multiple local optimal less than the difference.However,DARPSO can obtain better results when the values of global optimal and local optimal have larger difference.AQPSO and DARPSO algorithm improved QPSO algorithm,ARPSO algorithm and traditional PSO algorithm to some extent,especially for complex high-dimensional function optimization problems.