A model predictive control method for multi-energy flow of smart community combined with stochastic response surface method is proposed. In this method, the stochastic response surface method is adopted to obtain the distribution characteristics of forecasting errors of the distributed wind power outputs and photovoltaic power outputs, the load demand as well as the spot price, and the distribution curve of probability density is obtained and discretized. Then the roulette algorithm is used to get the initial scene set and the nearest neighbor ...