Energy storage sizing in plug-in Electric Vehicles: Driving cycle uncertainty effect analysis and machine learning based sizing framework

Abstract

Transportation electrification has been considered as one of the promising solutions towards clean environment. On-board Energy Storage System (ESS) has significant influence on the cost and reliability of Electric Vehicles (EVs). Furthermore, the choice of the ESS in terms of type and size, i.e., its capacity heavily depends on the driving patterns which, usually, vary significantly depending on factors like driver characteristics, geographical locations and traffic congestion. Hence, it is crucial to understand the effect of driving patterns/cycles on the sizing of ESS in order to avoid over/under-sizing. In this context, we investigate the effect of driving cycle uncertainties on the optimal sizing of ESS in EVs. We specifically focus on the uncertainty arising from the unpredictability of the future state of speed and acceleration and model the same as a scaling factor on velocity-magnitudes and time-windows of the velocity profile. Subsequently, (i) we derive the quantitative relationship between uncertainty and corresponding change in ESS size based on four thousand test driving cycles; (ii) we develop a framework which provides a guideline towards the choice of appropriate sizing of ESS in EVs subject to uncertainty in driving cycle; and (iii) we propose a machine-learning based framework that can enable a data-driven approach for ESS sizing with lower computational requirement. We choose a city bus application to illustrate our approach and consider three different ESS configurations: (i) Li-ion battery alone, (ii) Supercapacitor alone, and (iii) Battery–Supercapacitor hybrid ESS.