Learning Optimal Energy Management in Hybrid Vehicles
摘要
This chapter addresses the problem of power management in hybrid vehicles using the control design proposed in Chap. 7 . More precisely, the objective here is to minimize the energy used to travel on a specific path in the absence of a priori knowledge regarding the traffic conditions and the driver’s patterns. This leads to a very rich set of possibilities in terms of speed and acceleration profiles. In this study, real-life data instances are used which are collected on a specific path in Tokyo, Japan, in order to illustrate the efficiency of the so-designed feedback law that minimizes the overall energy used while guaranteeing a terminal state of charge of the vehicle’s battery. Given the specific problem at hand which stems from the quite long path to be considered for each instance of the uncertainty realization, specific problem-dependent steps are needed in order to accelerate the solution of the underlying deterministic Optimal Control Problem (OCP)Optimal control problem based on which, the uncertainty-aware feedback design proposed in Chap. 7 is built. This use-case shows another instance of a seemingly general rule according to which, efficient design cannot avoid, at least at some stage, the use of problem-dependent treatment and innovative thought.