Particle swarm optimisation with stochastic ranking for constrained numerical and engineering benchmark problems

Most of the real world science and engineering optimisation problems are non-linear and constrained. This paper presents a hybrid algorithm by integrating particle swarm optimisation with stochastic ranking for solving standard constrained numerical and engineering benchmark problems. Stochastic ranking technique that uses bubble sort mechanism for ranking the solutions and maintains a balance between the objective and the penalty function. The faster convergence of particle swarm optimisation and the ranking technique are the major motivations for hybridising these two concepts and to propose the stochastic ranking particle swarm optimisation (SRPSO) technique. In this paper, SRPSO is used to optimise 15 continuous constrained single objective benchmark functions and live well-studied engineering design problems. The performance of the proposed algorithm is evaluated based on the statistical parameters such mean, median, best, worst values and standard deviations. The SRPSO algorithm is compared with six recent algorithms for function optimisation. The simulation results indicate that the SRPSO algorithm performs much better while solving all the five standard engineering design problems where as it gives a competitive result for constrained numerical benchmark functions.