Totally disturbed chaotic Particle Swarm Optimization

Particle Swarm Optimization (PSO), classified as a swarm intelligence technique, mimics the well-informed swarming behavior of social species. A simple and effective searching strategy declares PSO as a potential member for solving various optimization problems. The present study embeds the concept of chaos at different stages of PSO, intending to enhance the convergence speed while trying to avoid stagnation and maintaining the solution quality. The proposed PSO variant is termed as “Totally disturbed PSO (TDPSO)”. The algorithm starts with a disturbed (chaotic) population, generated by considered chaotic system. Thereafter, when a certain number of iterations have elapsed and the searching process approaches equilibrium state, a relative velocity index is calculated for each particle to evaluate its present state and to decide whether or not the particle needs perturbation. The efficacy of proposed algorithm is tested against a set of benchmark problems and results are compared with existing Chaotic PSO and a standard PSO variant. Numerical results manifest that TDPSO works better over considered existing variants by effectively enhancing the searching capability and precision as well.