In this letter, we present a pollution source localization method combining particle swarm optimization (PSO) and artificial potential field (APF). PSO is used to control mobile agents to move toward the pollutant source according to concentration and position information. Meanwhile, to avoid collision during the movement, it is assumed that there exist repulsions around mobile agents and obstacles in monitoring area. APF is used to construct the corresponding repulsive force fields for each mobile agent. Besides, the residual energy is introduced to balance the energy consumption among mobile agents. Once the agent swarm locates the pollutant source, support vector regression is adopted to model the pollution diffusion based on concentration measurements and distribution of mobile agents. Experimental evaluations demonstrate the effectiveness and feasibility of our proposed method.