Abstract: An effective subarray partition method based on genetic algorithm (GA) is proposed to divide a large antenna array into contiguous subarrays with ununiform structure. In the genetic algorithm,the main-to-sidelobe ratio (MSR) of pattern is taken as fitness function. A subarray configuration with few obvious grating lobes is obtained by adding constraint condition in the genetic evolvement. A new method of beam scanning at subarray level is presented for the purpose of reducing the grating lobe by dividing the scanning region. In each scanning regional division, the beam scanning can be realized by beamforming at subarray level without adjustment of elements’ weights. In addition, the affection of different scanning angles on the array pattern is analyzed. To suppress the high level of sidelobes, subarray pattern has formed nulls in the directions of high sidelobes referring to adaptive array theory. Theoretical analyses and simulation results show that the main-to-sidelobe ratio of array pattern can be further improved, and also the range of scanning regional division is extended.