To address the shortages such as premature convergence and easily falling into local optimum of the bare bone particle swarm algorithm (BBPSO), we propose an improved BBPSO based on Von Neumann topology (VBBPSO). The new algorithm advocates taking lagging particles into account, constructs neighboring areas by applying the Von Neumann topology, replaces the global optimal solution with the neighboring optimal solution, introduces the central adjustment coefficient, and adjusts the evolutional central term and the discrete control item of the BBPSO algorithm in neighboring areas, thus improving the global exploration ability and the local development ability. Experimental results indicate that the performance of the VBBPSO has a better performance in comparison with other classical algorithms.