Abstract: Typical operation curve mining of energy storage system（ESS） can provide an overall cognition of power features of ESS and provide a calculation basis for determining capacity configuration in a certain application scenario. Based on restraining PV power fluctuation application scenario, a typical operation curve mining algorithm based on cloud model of ESS was proposed in this paper. Frequency distribution of longitudinal moment power of ESS was decomposed into several normal cloud groups by multiple cloud transform, typical power values was calculated by expectations of normal cloud model group with weight sum. In this way, the typical operation curve mining of ESS was obtained in integration. Take a nonparametric kernel density estimation for ESS power distribution under 6 typical weather patterns and the typical operation curve, the fitness bias was obtained to verify the rationality of the typical operation curve. ESS capacity configuration was obtained according to the typical operation curve. Effectiveness and correctness of the typical operation curve was verified by a simulation example. Theoretical support of ESS capacity configuration was provided in this paper.