In this paper is proposed a new method for optimal allocation and sizing of Photovoltaics (PVs) systems in power distribution network to reduce losses in feeders and minimize the economic aspects of the PVs implementation. The proposed technique, called Modified Particle Swarm Optimization with Dynamic Momentum (MPSO-DM) is proposed so that can be applied in distribution system, that requires a high number of decision variables in the optimization process. This becomes necessary because the classical Particle Swarm Optimization (PSO), in general, presents a performance drop in the exploration of the search space when applied to these situations. The proposed method was tested in a real distribution network of the Federal University of Paraiba (79 buses), which results show a improvement in performance, yielding better solutions when compared to the traditional Modified Particle Swarm Optimization (MPSO) method and MPSO with linear decrement inertia weight.