Energy prediction and management strategy of extended range electric vehicle based on fused multi-source information

To achieve the accurate energy estimation and improve the economy performance of Extended Range Electric Vehicle (EREV) under various information from vehicle-road-cloud systems, an energy consumption prediction and optimal control strategy for EREV is proposed based on fused average vehicle speed, road segment length, and traffic congestion information. In the dataset, the average speed of different road segments, segment length, and congestion information are selected as the three neurons in the input layer of Back Propagation (BP) neural network model. Through iterative gradient search training, the network parameters are determined, with the model output set as the energy consumption of future road segments. Real-time traffic information is obtained, and the predicted energy consumption output by the BP model is utilized to dynamically adjust the State of Charge (SOC) balance point. The operating point of the range extender engine is optimized in real time to achieve optimal energy efficiency. The Simulations results indicate that when the numbers of predicted road segments are 4, 5, and 6, the predictive strategy enables the pre-estimation of energy consumption levels for future road segments. By reasonably adjusting SOC balance point, energy replenishment is provided in high-efficiency zones of the range extender for congested segments.