A novel PSO with piecewise-varied inertia weight
Quande Qin, Li Li, Rongjun Li · 2010
To choose the appropriate value of inertia weight can improve the performance of PSO by means of making a good balance between exploration and exploitation in search process. This paper presents a novel inertia weight variation method based on a piecewise function, in which there are two parts: one is nonlinear decreasing to enhance the explorative ability; the other is linear decreasing just as standard PSO algorithm. The two key parameters in the proposed PSO algorithm (PSO-PIW) are identified through two experimental simulations. The results of several benchmark functions tested demonstrate good performance of PSO-PIW in solving multimodal problems compared with some other inertia weight variation methods in PSO.