Chaotic multi-swarm particle swarm approach for solving numerical optimization problems

摘要

Different fields of study are faced with several optimization problems which can either be discrete, nonlinear, linear, continuous, non-smooth, or non-convex in nature. The continuously differentiable problems can be handled using several conventional methods such as the gradient-based methods, but such methods may not be ideal for the complex problems such as the non-convex or non- differentiable prob-lems. Despite the existing number of methods for solving complex optimization problems, achieving optimal results is still difficult without much computational effort and cost input. The Particle Swarm Optimization (PSO) algorithm is a common optimization algorithm which is still suffering from an unbalanced local search (exploitation) and global search (exploration). The Meeting Room Approach (MRA) was recently developed as a multi-swarm model which for enhancing the exploration and exploitation in the PSO algorithm. In proposed Multi-swarm approach, the algorithm starts from a uniformly generated positions, which may start from not good positions. In other words, the algorithm may have a slow convergence due to the initial positions. In this paper, a Logistic map was used to initiate a multi-swarm PSO to enable it to start from better positions. The performance of the proposed algorithm was evaluated on several numerical optimization problems and its convergence was found to be faster compared to the original model.