Improved multidisciplinary multiobjective collaborative optimization technique based on NSGA-II algorithm

Abstract

The multiobjective collaborative optimization (MCO) has been widely adopted in concurrent engineering design as a good multiobjective optimization approach provided the problem of optimization results converging often to a local extremum has been taken care of. It has been demonstrated to yield useful solutions to a wide range of optimization problems under concurrent condition both technical and non-technical ones. A solution strategy for MCO that overcomes this drawback is proposed here on the grounds of genetic NSGA-II algorithm. It is an adaptive heuristic search technique that controls the distances between the individual and system's feasible region during genetic evolution. In the evolving NSGA-II based non-dominated sorting, the individual threshold of infeasibility degree is reduced gradually via genetic evolution and the individual's feasibility capacity is determined by its infeasibility degree and the threshold. It has been applied to a benchmark example of speed reducer design, and the obtained results demonstrate the proposed technique is comparably more efficient and may offer substantial advantages over the gradient-based optimization methods.