A New Accelerated Multi-objective Particle Swarm Algorithm. Applications to Truss Topology Optimization

We propose a new algorithm of computation using particle swarm in order to solve multi-objective problems more quickly and e ectively. This approach, called accelerated multi-objective particle swarm, is partially based on our previous work [4] and incorporates a vector function as objective function and it uses matrix computation to develop the Pareto front. Unlike all these studies which use inertia weight to develop Pareto front and an external archive to save non-dominated solution, we will modify this algorithm for causing it to use matrix computation, then this algorithm incorporates function vector as objective function and uses Pareto dominance for selecting best solutions and updating Pareto set. In addition, we also propose a new strategy of initialization that contributes too to the acceleration of the algorithm. The resulting algorithm is applied to multi-objective topology optimization of truss structures. The results produced by such a strategy illustrate that the algorithm is competitive with NSGA-II and MISA in terms of converging to the true Pareto front. It maintains the diversity of the population, generates better trade-o s and demonstrates that the matrix computation PSO can be used as a reliable numerical optimization tool.