Feedback learning particle swarm optimization

Abstract

In this paper, a feedback learning particle swarm optimization algorithm with quadratic inertia weight (FLPSO-QIW) is developed to solve optimization problems. The proposed FLPSO-QIW consists of four steps. Firstly, the inertia weight is calculated by a designed quadratic function instead of conventional linearly decreasing function. Secondly, acceleration coefficients are determined not only by the generation number but also by the search environment described by each particle’s history best fitness information. Thirdly, the feedback fitness information of each particle is used to automatically design the learning probabilities. Fourthly, an elite stochastic learning (ELS) method is used to refine the solution. The FLPSO-QIW has been comprehensively evaluated on 18 unimodal, multimodal and composite benchmark functions with or without rotation. Compared with various state-of-the-art PSO algorithms, the performance of FLPSO-QIW is promising and competitive. The effects of parameter adaptation, parameter sensitivity and proposed mechanism are discussed in detail.