Abstract: In view of the problem that during engineering optimization the optima generally are located at near constraint boundaries due to the limitation of resources, but current constrained optimization algorithms tend to focus on handling constraint mechanism and seldom to search near constraint boundaries, an algorithm of two-step constrained particle swarm optimization (PSO) algorithm is proposed in this paper. In the first step, PSO algorithm based on penalty function is adopted for optimization with a pointer for speed reset used to prevent optimization from falling into stagnation. In the second step, subset constraint boundary reduction equation is employed to acquire information of constraint boundaries, and sequential quadratic programming (SQP) is utilized to conduct local search on boundaries. Finally the results of two steps are compared and the better one is chosen as the global optimum. As an application, the modified algorithm proposed is used to perform body lightweight optimization on a fuel cell sedan under both side collision and roof crash conditions, resulting in a 10.92% mass reduction of some body panels taking part in optimization while ensuring the structural crashworthiness of the vehicle.

Key words: car-body lightweight design, particle swarm optimization algorithm, constrained optimization, subset constraint boundary reduction equation, boundary local search