The invention discloses an optimization method for a mixed energy system based on wind-solar complementation under a random model. In reference to historic data, an ARMA model is established and is more accurate compared with a traditional mathematical model; stability index EENS is introduced, so that the stability of the overall model is improved; an Fischer-Burmeister optimization method is used in combination with SAES, and in offspring individuals generated by mutation, the individuals with high adaptability will be limited within a restrained range specified by the Fischer-Burmeister optimization method. Compared with a traditional method, the optimal result can be solved under the overall situation, the condition of local optimum is avoided, meanwhile, the calculation amount is greatly reduced, and the speed during optimization is increased.