In order to improve the efficiency of distributed systems over a network, various distributed optimization algorithms have been developed recently. In particular, for optimization problems with convex and differentiable functions, sophisticated algorithms have been proposed, motivated by energy network systems such as smart grid. As an algorithm with easier implementation and wider range of applications, this paper proposes a distributed optimization algorithm that can deal with optimization problems with nonconvex and non-differentiable functions by combining a particle swarm optimization algorithm and an average consensus algorithm. Moreover, the convergence property of the proposed algorithm is proven under mild assumptions. Through numerical experiments, the effectiveness of the proposed algorithm is illustrated.