Optimization configuration of photovoltaic-storage system capacity based on non-parametric kernel density estimation

Energy storage technology is helpful to reducing the forecast errors between the real power outputs and the forecasting power outputs at a photovoltaic station and improving the reliability of the forecasting power outputs as the references of the power system dispatch. Consequently, a capacity allocation method of photovoltaic-storage system based on the forecasting error distributions of photovoltaic power outputs was proposed. A combination forecasting model of Markov chain and continuous method was adopted to realize the photovoltaic power short-term prediction, relying on that the storage system could compensate for the forecasting errors between the actual photovoltaic power and predictive power. The prediction error distribution model was established by the non-parametric kernel density estimation method in which the total capacity configuration of energy storage system was calculated according to the maximum energy absolute value of the cumulative distribution function. The simulation analysis of a 20 MW (peak value) photovoltaic power station based on the measured data reveals that, with the capacity demand meeting rate 95%, capacity configuration of energy storage system is 9 MWh. This method provides a theoretical basis for the rational allocation of energy storage system for PV power station, the photovoltaic power station has the better adjustable and controllable ability. ©, 2015, Science Press. All right reserved.