Connected PHEV Energy Management based on Global Driving Cycle Construction

In this paper, a global vehicle driving cycle construction method is newly proposed, to enhance the energy management performance of a connected plug-in hybrid electric vehicle (Connected PHEV). We propose a three-step novel real-time future driving cycle construction method. First, historical driving cycles are collected and each of them is divided into a number of speed segments to form a database. Artificial neural network (ANN) is employed to learn the nonlinear correlation between the key features of adjacent speed segments along the entire trip. Finally, this trained ANN model is deployed in real-time to predict the next most possible speed segment based on current driving condition of the vehicle. By sequential operating, the global driving cycle can be constructed. The method is validated in a fixed-route city bus driving scenario using real-world data. Model predictive control (MPC) are adopted to solve the energy management problem. Simulation results illustrate that the driving cycle construction method is able to improve the fuel economy of PHEV by over 29% compared with traditional energy management method.