Battery Aging-Aware Online Optimal Control: An Energy Management System for Hybrid Electric Vehicles Supported by a Bio-Inspired Velocity Prediction

Abstract

In this manuscript we address the problem of online optimal control for torque splitting in hybrid electric vehicles that minimises fuel consumption and preserves battery life. We divide the problem into the prediction of the future velocity profile (i.e. driver intention estimation) and the online optimal control of the hybrid powertrain following a Model Predictive Control (MPC) scheme. The velocity prediction is based on a bio-inspired driver model, which is compared on various datasets with two alternative prediction algorithms adopted in the literature. The online optimal control problem addresses both the fuel consumption and the preservation of the battery life using an equivalent cost given the estimated speed profile (i.e. guaranteeing the desired performance). The battery degradation is evaluated by means of a state-of-the-art electrochemical model. Both the predictor and the Energy Management System (EMS) are evaluated in simulation using real driving data divided into 30 driving cycles from 10 drivers characterised by different driving styles. A comparison of the EMS performances is carried out on two different benchmarks based on an offline optimization, in one case on the entire dataset length and in the second on an ideal prediction using two different receding horizon lengths. The proposed online system, composed by the velocity prediction algorithm and the optimal control MPC scheme, shows comparable performances with the previous ideal benchmarks in terms of fuel consumption and battery life preservation. The simulations show that the online approach is able to significantly reduce the capacity loss of the battery, while preserving the fuel saving performances.