Novel inertia weight strategies for particle swarm optimization

Abstract

The performance of PSO algorithm depends greatly on the appropriate parameter selection strategies for fine tuning its parameters. This paper proposes three new nonlinear strategies for selecting inertia weight which plays a significant role in particle's foraging behaviour. The PSO variants implying these strategies are named as: fine grained inertia weight PSO (FGIWPSO); Double Exponential Self Adaptive IWPSO (DESIWPSO) and Double Exponential Dynamic IWPSO (DEDIWPSO). In FGIWPSO, inertia weight is obtained adaptively, depending on particle's iteration wise performance and decreases exponentially. DESIWPSO and DEDIWPSO employ Gompertz function, a double exponential function for selecting inertia weight. In DESIWPSO the particles' iteration wise performance is fed as input to the Gompertz function. On the other hand DEDIWPSO evaluates the inertia weight for whole swarm iteratively using Gompertz function where relative iteration is fed as input. The efficacy and efficiency of proposed approaches is validated on a suite of benchmark functions. The proposed variants are compared with non linear inertia weight and exponential inertia weight strategies. Experimental results assert that the proposed modifications help in improving PSO performance in terms of solution quality as well as convergence rate. © 2013 Springer-Verlag Berlin Heidelberg.