A novel method for improving multi-objective evolutionary algorithms is described in the paper, which can simultaneously optimize nonuniform conformal subarrays partitions and conformal subarrays amplitude excitation. For the purpose, the new multi-objective functions have been designed, double alternative genetic operator in the improved strength Pareto evolutionary algorithm (SPEA2) is used to improve searching efficiency and convergence and radiant characteristic of the whole array. In system simulation, considering project application, the proposed algorithm is employed to divide a 20*20 array into 25 nonuniform subarrays and each subarray amplitude excitation is optimized. Simulation results show that the performance of peak sidelobe level (PSLL) and beamwidth in azimuth and elevation of antenna array in scan airspace is improved greatly. It has been proved that this is effective for this algorithm to improve radiant characteristic of the whole array