In this paper, a novel variant of particle swarm optimization (PSO), named memetic particle swarm optimization algorithm (MeSwarm), is proposed for tackling the overshooting problem in the motion behavior of PSO. The overshooting problem is a phenomenon in PSO due to the velocity update mechanism of PSO. While the overshooting problem occurs, particles may be led to wrong or opposite directions against the direction to the global optimum. As a result, MeSwarm integrates the standard PSO with the Solis and Wets local search strategy to avoid the overshooting problem and that is based on the recent probability of success to efficiently generate a new candidate solution around the current particle. Thus, six test functions and a real-world optimization problem, the flexible protein-ligand docking problem are used to validate the performance of MeSwarm. The experimental results indicate that MeSwarm outperforms the standard PSO and several evolutionary algorithms in terms of solution quality.