An improved partial swarm optimization (IPSO) algorithm has been presented to overcome the defect of trapping in local optimum in the PSO algorithm. In the IPSO, swarm A is selected to do particle swarm optimization (PSO), and the maximum number of iterations should be divided into several stages, a new swarm B will be generated randomly outside of swarm A according to the constraints on specific issues at the end of each stage, and the swarm B also obey the PSO. In the next iteration stage of swarm A, the better result through comparison of the swarm B and A will be chosen to update the current optimization result of swarm A. Then swarm A continues to calculate, after repeated comparison to find the optimal solution. The IPSO is applied to the optimization of Rectifier Transformer designation. The feasibility and validity of the proposed method are proved by the contrast between handwork results and IPSO results.