Game-Theoretic Approach for Complete Vehicle Energy Management

Abstract

This paper describes a game-theoretic approach and the calculation of an online implementable strategy for solving the Complete Vehicle Energy Management (CVEM) problem, which aims at minimizing the global fuel consumption of a hybrid heavy-duty truck including all its auxiliary systems. The approach is based on a two-level single-leader multi-follower game, in which the driver is considered as a leader and each controlled auxiliary is considered as a follower. In the first level, a sequential game between the leader (driver) and each follower (auxiliary) is played and the corresponding Stackelberg strategy is computed offline and stored in a lookup table. In the second level, a simultaneous game is played among all followers and an online iterative process is introduced to find an approximation of a Nash equilibrium for all followers. This approach is tested on a hybrid heavy-duty truck model where a high-voltage battery and an electric refrigerated semi-trailer are considered. The performance in terms of fuel economy is found close to the true optimal solution.