Modified Particle Swarm Optimization

Particle Swarm Optimization (PSO) is a very popular optimization technique, but it suffers from a major drawback of a possible premature convergence i.e. convergence to a local optimum and not to the global optimum. This paper attempts to improve on the reliability of PSO by addressing the drawback. This problem of premature convergence is more probable with the problems, which have the global optimum surrounded by the positions returning bad function values as these regions remain unexplored and swarm can get stuck into some local optimum. In the present paper, a modified particle swarm optimization is proposed to address this problem. During each iteration cycle, while deciding new positions, some particles will be chosen to give weightage to the worst solutions instead of good solutions. It will enable them to exploit the region for a probable global optimum. This modified method would free PSO from local optimum solutions; enable it to progress towards the global optimum searching over wider area. So the probability, of not getting trapped into local optima gets enhanced which gives better assurance to the achieved solution. Numerical experiments on the benchmark functions have been discussed in the paper.