Configuration Optimization of Hybrid Energy Storage System Based on Dynamic Weighted Power Prediction Error Distribution

In order to quantify the impact of wind and photovoltaic (PV) power volatility on Wind-PV-Energy storage system sizing, the optimal capacity configuration is investigated, focusing on dynamic power prediction distribution features. Firstly, a dynamically weighted Nonparametric Kernel Density Estimate-Gaussian Mixture Model (NKDE-GMM) algorithm is proposed. Based on the above distribution algorithm, a Wind-PV-Energy storage capacity optimization model based on improved Artificial Fish Swarm Algorithm (AFSA) is applied. The objective of this model is to minimize the life cycle cost of the energy storage system, considering the penalty cost associated with the uncertainty of wind and PV energy prediction power. Finally, the optimization configuration outcomes at different confidence levels are compared by the proposed comprehensive evaluation index (CEI). Case study shows that minimum and maximum values of CEI are at 95% and 90% confidence level, respectively. Furthermore, CEI increases by 0.067 at the 90% confidence level compared to the one at 95% confidence level.