Predictive Multi-Objective Operation Strategy Considering Battery Cycle Aging for Hybrid Electric Vehicles 2018-01-1011

Due to the new CO₂ targets for vehicles, electrification of powertrains and operation strategies for electrified powertrains have drawn more attention. This article presents a predictive multi-objective operation strategy for hybrid electric vehicles (HEVs), which simultaneously minimizes the fuel consumption and the cycle aging of traction batteries. This proposed strategy shows better performance by using predictive information and high robustness to inaccuracy of predictive information.

In this work, the benefits of the developed operation strategies are demonstrated in a strong hybrid electric vehicle (sHEV) with P2-configuration. For the cycle aging of a lithium-ion battery, an empirical model is built up with Gaussian processes based on experimental data. Two different optimization algorithms “Deterministic Dynamic Programming” (DDP) and extended “Multi-Objective Equivalent Consumption Minimization Strategy” (MO-ECMS) are carried out with a priori knowledge of cycle information to obtain the Pareto front between fuel consumption and battery cycle aging. In Worldwide harmonized Light vehicles Test Cycle (WLTC), halved battery cycle aging leads to 4% more fuel consumption compared with the original Equivalent Consumption Minimization Strategy (ECMS).

In order to achieve the maximal potential of the multi-objective operation strategy for in-vehicle optimization, two variants of predictive Multi-Objective Equivalent Consumption Minimization Strategy (pMO-ECMS), i.e., adaptive and causal MO-ECMS, are further developed based on the MO-ECMS. Different methods are considered to incorporate predictive information into the operation strategies. For example, the navigation-based road information is used to modify parameters of the pMO-ECMS; a reference State of Charge (SoC) trajectory is generated with estimated vehicle speed and road slope. The performance of the pMO-ECMS with respect to the MO-ECMS is evaluated with ideally and roughly estimated speed profile. In further development, the simulation results will be validated in an experimental vehicle.