Minimizing Mathematical Benchmark Test Functions using Selection Based Particle Swarm Optimization

Abstract

This article is based on a new technique called selection based particle swarm optimization (SBPSO) which has been developed and implemented to mathematical benchmark functions. In SBPSO, the numbers of particles are decreased in each iteration by decrement factor ‘t’. Elimination of particles is based on their function value. SBPSO is used as a guideline for performance testing by different functions. The benchmarks we used are Rosen brock, Beal and booth. We implemented the original algorithm and obtained optimized results for every function. The results of SBPSO have been compared with that of BPSO, and SBPSO is found to converge faster with more accurate results. SBPSO is compared with BPSO by the number of times the objective function has been processed (counted) in optimization technique.