In high-power wind generation system, grid-side LCL filter is usually applied to suppress high frequency harmonics generated by the converter. But present methods for parameters design of LCL filter have complex computation, low accuracy and poor performance. This paper proposes an optimized design method based on adaptive genetic algorithm for LCL filter. It uses adaptive crossover and mutation probability to increase the evolutionary efficiency and decrease computation time. In addition, the design constraints and filter performance are considered comprehensively. At last, the PSCAD/EMTDC simulation results indicate that the optimized filter has favorable ability of harmonic suppression, and it is significant for practical application.