Memetic multi-topology particle swarm optimizer for constrained optimization

During the last two decades, a considerable number of particle swarm variants have been introduced. However, no single variant consistently performed well over a range of test problems with different mathematical properties. In this paper, a memetic multi-topology particle swarm optimizer (MMTPSO) is introduced for solving constrained optimization problems. MMTPSO utilizes the strengths of two different particle swarm topologies and during the evolution process the algorithm is designed to emphasize the best performing topology. Moreover, to increase the convergence pattern of the proposed algorithm, a local search algorithm is periodically used. MMTPSO shows a superior performance to its independent variants, as well as other state-of-the-art algorithms, by solving 13 well-known test problems.