Optical phased array LiDAR systems possess small sizes and high-speed scanning capabilities. However, their performance is severely limited by the grating lobe problem. The non-equal spacing optical phased array method resolves this limitation by breaking the periodicity of optical phased arrays and suppressing grating lobe formation. The proposed particle swarm optimization algorithm improved upon the traditional algorithm by introducing adaptive parameter optimization, a perturbation strategy, and an optimal retention strategy to improve the optimization effect of the traditional algorithm. Moreover, it continuously optimized the optimal element distribution of optical phased arrays to obtain the element spacing values that reduced the level of grating lobes. The grating lobe optimization method based on the improved particle swarm algorithm was simulated and tested. The test results revealed that the maximum grating lobe value can be reduced to 0.0968, effectively addressing the grating lobe problem. Finally, the proposed algorithm was used to study the two-dimensional rectangular planar array and an optimization design method for a two-dimensional nonuniform optical phased array was proposed.