Multi-objective particle swarm optimization algorithm and its application to optimal design of tolerances

A Multi-objective Particle Swarm Optimization (MOPSO) algorithm is proposed to acquire the Pareto solution set of multi-objective problems by merging the effective Particle Swarm Optimization (PSO) algorithm and the techniques to deal with the multi-objective problems. In MOPSO, the particles are redefined according to the conceptions of Pareto optimality, a fast non-dominated sorting approach is given and a new mechanism of local memorization and global information sharing is also proposed. Theoretical analysis shows that the computation complexity of the MOPSO algorithm is less than that of some traditional evolutionary algorithms. When the MOPSO is applied to solve optimal design of component tolerances, a typical engineering problem, the results show that this new algorithm has high efficiency and can get the Pareto optimal set in one run. The computation results additionally reveal and prove several important rules in design of tolerances.