Answers are grounded in this paper and cite their sources.
As the electrification of public transport and the adoption of variable renewable energy accelerate the transition to carbon neutrality, integrating local photovoltaic (PV) surplus power into electric bus charging operations becomes increasingly critical. However, uncertainties in PV generation and traffic delays often reduce the effective utilization of PV surplus due to missed charging opportunities. To address these challenges, this study proposes a dynamic charging scheduling method based on model predictive control (MPC), which adaptively updates the schedule using quasi-real-time, district-scale information. The framework integrates real-time traffic delays in the General Transit Feed Specification format (GTFS Realtime), smart meter measurements, and meteorological satellite observations—data sources currently available in real cities. At each update step, the system forecasts PV surplus power using a machine learning model that captures temporal weather conditions and localized PV surplus trends around charging stations, while detecting bus delays at each station. Based on this information, the optimal charging schedule is updated every 30 min to adaptively maximize PV surplus utilization. Numerical experiments simulating an entire year demonstrate the effectiveness of the proposed method. Compared to a fixed day-ahead schedule and a rule-based charging method, it improves the annual average PV surplus utilization rate by up to 11.9% and reduces annual average grid power purchases by up to 15.6%. These results highlight the potential of combining MPC with quasi-real-time, district-scale data to proactively and robustly integrate renewable energy into public electric bus operations under uncertainty.
No takes yet. Share an insight, caveat, or question.
Osaki et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: