To improve the beam collection efficiency (BCE) of the microwave wireless power transmission (MWPT) system while reducing the peak sidelobe level outside the receiving area (CSL) and system cost, this paper proposes a new subarray partition technique and a nonuniform sparsely distributed quadrant symmetric planar array (NSDQSPA) model. A particle swarm optimization algorithm based on multiple-objective with nonlinear time-variant inertia and learning factor improved particle swarm optimization (MO-NTVILF-IPSO) is also proposed. The one-step multi-objective subarray partition algorithm adopts dynamic weight and dynamic learning factor to carry out one-step optimization on the array element arrangement of the transmitting array. The optimization algorithm simultaneously optimizes two performance indicators: the ∆BCE, which represents the optimization accuracy for the BCE, and the α ref , which represents the mean square error of the excitation amplitude before and after the subarray partition. Many simulation results show that the BCE is 94.91%, and the CSL is −13.41 dB when the transmitting array with an aperture of 4.5λ × 4.5λ is divided into six subarrays. The simulation results further demonstrate that the proposed subarray division method is appropriate for the MWPT system and that the algorithm in this paper, when the array elements with the same excitation amplitude are divided for the planar transmitting array on the array model, and can guarantee relatively high BCE and relatively low complexity of the system feed network.