A Novel Multiobjective Particle Swarm Optimization Algorithm with Dynamic Resource Allocation

This paper proposes a novel multiobjective particle swarm optimization algorithm with dynamic resource allocation, showing promising performance especially for tackling some complicated multiobjective optimization problems. With the decomposition approach, each particle is assigned to optimize one subproblem with a novel velocity update strategy to speed up the convergence. Moreover, a dynamic resource allocation strategy is designed based on the relative improvement of subproblems, which can reasonably allocate computational resource to the particles that are able to search superior solutions. By this way, the proposed algorithm not only has strong exploratory capability, but also can converge quickly to the true Pareto-optimal front. The experimental results fully demonstrate the superiority of our proposed algorithm over four state-of-the-art multiobjective optimization algorithms, when tackling thirty-five test problems.