Abstract: Since using the traditional particle swarm optimization (TPSO) algorithm simultaneously to optimize the form of all 3D array elements takes a long time and is easy to converge to a local result.An improved particle swarm optimization (IPSO) algorithm based on the parallel computing idea is presented in this paper.The multiple particle swarms are initialized simultaneously,and the dimension reduction of each particle is achieved by means of optimizing the parameters of an array arm (optimized arm) only.Subsequently,by using the linearly decreasing inertial weight to pre-optimize the multiple particle swarms at the same time,the intermediate solution is obtained to form the new particle with regained dimension.The new particle swarm optimization process continues with the least inertia weight to output the solution until the stop condition is satisfied.A 5-arm star-shaped 3D array is optimized and designed in this paper,it is found that the computing time is reduced and it can obtain a better result for the IPSO algorithm.At the end of this paper,the six simulation tests are used to discuss the characteristics of the arrays designed in this paper.