With the increasing use of electrical equipment, it is necessary to build new substation to expand the scale of power system. Combining the advantages of KH algorithm and PSO algorithm, a novel hybrid algorithm, called KH–PSO, is presented for the application of substation site selection. In the new improved KH–PSO, the randomly generated initial population is divided into two subpopulations for the krill algorithm and the particle swarm algorithm. By population separation and merging, all individuals can exchange position information with each other. This method can avoid the premature convergence, and find the global optimal solution without introducing additional operators to the basic KH and PSO algorithms. To verify its performance, experiments are carried out on the problem of substation site selection. Based on the results, we can easily infer that the hybrid KH–QPSO is more efficient and accurate.