Data-driven predictive energy management and emission optimization for hybrid electric buses considering speed and passengers prediction

Abstract

The energy-saving and emission reduction potential of hybrid electric vehicles are of great significance to the environment’s sustainable development. The trade-off between energy consumption economy and environmental friendliness is essential. To promote efficiency while reducing the emission of hybrid electric buses (HEB), we propose a novel predictive energy management strategy with passenger number prediction and exhaust emission optimization. An integrated prediction of vehicle speed and passenger number based on deep learning is proposed to predict future power demand accurately. An emission penalty is introduced into the objective function. Then, the impacts of passenger number prediction and the penalty on energy consumption and exhaust emissions are discussed. Simulation results show that the deep neural network predictor performs better in predicting speed and passenger number than Markov chain and radial basis function neural network predictors. The proposed energy management’s energy efficiency reaches 97.02% of global dynamic programming and 2.49% higher than that of instantaneous optimal control. With the exhaust emission optimization, CO2, CO, NOx, and HC emissions are reduced by 6.22%, 10.51%, 6.3%, and 4.83%, respectively, while the energy consumption cost is only increased by 1.34%. The proposed approach is verified to be environmentally friendly and energy-saving.