A Multi-Timescale Joint Stochastic Optimization Scheduling Method for PV-Storage Charging Stations under Correlated Uncertainties

Abstract

: To address the scheduling challenges caused by uncertainties in both photovoltaic (PV) generation and charging demand at solar-powered stations, a stochastic multi-timescale scheduling approach is introduced for PV-storage charging stations. In the day-ahead scheduling phase, a Copula-based joint scenario generation and reduction method is employed to account for the assumed statistical dependence between PV output and electric vehicle (EV) charging loads. Based on the reduced representative scenarios, the initial state of charge (SOC) of the energy storage system (ESS) is optimized to minimize the expected daily operating cost while accounting for correlated source-load uncertainty. In the intraday scheduling stage, predictions are updated every 15 min using real-time data. The ESS charging and discharging power is updated through 15-min rolling optimization to respond to updated source-load information and time-of-use electricity prices. This is achieved through rolling optimization, which adopts a hybrid of Simulated Annealing and Particle Swarm Optimization (SA-PSO) algorithms. Numerical results indicate that the proposed multi-timescale framework reduces operating costs compared with the case without ESS while accounting for the assumed PV-EV dependence structure. The framework also mitigates grid-power fluctuations under the considered numerical assumptions.