An improved particle swarm optimization with decline disturbance index (DDPSO) for multi-objective job-shop scheduling problem

As same with many evolutional algorithms, performance of simple PSO depends on its parameters, and it often suffers the problem of being trapped in local optima so as to cause premature convergence. In this paper, an improved particle swarm optimization with decline disturbance index (DDPSO), is proposed to improve the ability of particles to explore the global and local optimization solutions, and to reduce the probability of being trapped into the local optima. The correctness of the modification, which incorporated a decline disturbance index, was proved. The key question why the proposed method can reduce the probability of being trapped in local optima was answered. The modification improves the ability of particles to explore the global and local optimization solutions, and reduces the probability of being trapped into the local optima. Theoretical analysis, which is based on stochastic processes, proves that the trajectory of particle is a Markov processes and DDPSO algorithm converges to the global optimal solution with mean square merit. After the exploration based on DDPSO, neighborhood search strategy is used in a local search and an adaptive meta-Lamarckian strategy is employed to dynamically decide which neighborhood should be selected to stress exploitation in each generation. The multi-objective combination problems with DDPSO for finding the pareto front was presented under certain performance index. Simulation results and comparisons with typical algorithms show the effectiveness and robustness of the proposed DDPSO.