Dynamic charging demand forecasting and flexible scheduling of photovoltaic microgrid in traffic congestion scenario

Abstract Under the traffic congestion scenario, the charging demand of electric vehicles has significant temporal and spatial fluctuations, which pose higher requirements for the operation and scheduling of optical storage and charging microgrids. Therefore, aiming at the dual uncertainties of traffic congestion and PV output, this study proposes a collaborative optimization method by integrating dynamic demand forecasting and time-sharing elastic scheduling. Firstly, an ultra-short-term prediction model of photovoltaic power generation based on Variational Mode Decomposition-Long Short-Term Memory (VMD-LSTM) and an error compensation mechanism is established to realize high-precision prediction of photovoltaic output under different weather conditions. Secondly, by combining a traffic flow dynamic model and user behavior analysis, a charging demand prediction model under congestion scenarios is constructed. On this basis, a time-sharing electricity price optimization and energy storage coordinated dispatching model is also established with the dual objectives of maximizing the total revenue of the microgrid and customer satisfaction, which is solved by the dynamic programming method. The experimental results indicate that the mean absolute percentage error (MAPE) of the proposed PV prediction model is 2.72% under various weather conditions, which is better than that of VMD-LSTM (2.84%), SVM (6.22%), and LSTM (4.76%). Under three typical congestion scenarios, the system revenue increases by about 15.8% and 18.7% respectively through dynamic adjustment of electricity price and optimal scheduling of energy storage, while the customer satisfaction only decreases by about 1%. Obviously, the proposed strategy is very effective in improving economy and maintaining good user experience. The main innovations lie in that a photovoltaic ultra-short-term forecasting model integrating VMD-LSTM and an error compensation mechanism is proposed, a charging demand forecasting model considering the dynamic response of traffic congestion is constructed, and a time-sharing rolling optimization framework is designed to realize the coordinated scheduling of photovoltaic, energy storage, and congestion.