On the basis of the low power consumption and severe fluctuation load of residential area in microgrid, a model of maximum Lyapunov exponent via the phase space reconstruction is constructed combined with chaos theory. This method does not directly consider the impact of factors such as climate, electricity price, thus it requires less input and parameters. It utilizes C -C method to solve the delay time and the embedding dimension, and uses the improved maximum Lyapunov exponent method to forecast. The model is applied to forecast actual load data of a certain community in Anhui province. According to the time of use (TOU), a relative error indicator for time period based on the control of microgrid energy storage charge and discharge system is proposed. Simulation results demonstrate the efficiency and practicality of the algorithm.