Battery Health Aware Energy Management Strategy for Hybrid Electric Vehicle using Artificial Neural Networks

Abstract

Battery health aware energy management strategy (EMS) is designed for a power-split hybrid electric vehicle (HEV) by using Artificial Neural Networks (ANN). Three different speed profiles are selected to obtain a dataset. WLTP is used as the dynamic, transient speed profile, HWFET is used as the highway speed profile and NEDC is used as the third speed profile. Vehicle is operated in charge-sustaining mode in these drive cycles by using Equivalent Consumption Minimization Strategy (ECMS). Three different initial State-of-Charge (SOC) values are used in simulations. Each cycle is run for three different initial SOC values. Two ANN controllers are designed to control ICE torque and speed. Torque demand of the vehicle, SOC, and battery capacity fade are selected as the inputs of the ANN. The goal of this study is the investigation of battery degradation and fuel consumption by using ANN. Results for WLTP show that capacity fade can be reduced up to 14.85% and fuel consumption can be reduced 3.83% for the lowest initial SOC value. For intermediate initial SOC value, capacity fade is reduced by 13.80% and fuel consumption is reduced by 1.84%.  For highest initial SOC value, capacity fade is reduced by 14.70 % with the 5.75% increase of fuel consumption. Results are consistent for the other two drive cycles. Battery degradation is also reduced in HWFET and NEDC.