The fusion of multi-observation system can improve the estimation accuracy of target location, but there is no better solution to improve the efficiency of information fusion. To solve this problem, a distributed multi-population parallel genetic algorithm is presented by cooperative evolution. The algorithm divides observation system into multiple independent and parallel evolutionary sub populations. By setting a discrete fitness function, the sub population converges to an optimal value region, which can provide more information for the target population measurement fusion to effectively improve the fusion accuracy with the migration of individuals. The simulation results show that, in comparison with the genetic algorithm of centralized fusion and the parallel Chan algorithm of distributed fusion, the proposed algorithm can obtain better information fusion effect and a higher positioning accuracy.