A game theoretic approach to control of a hybrid electric vehicle powertrain.

摘要

The automotive community has long dealt with concerns over the environmental impact of personal transportation methods. Such concerns, coupled with political concerns such as gasoline prices, and increasing dependence on foreign oil has increased the research effort into alternative concepts for vehicle powertrains. One of the most prominent technologies is the hybrid electric vehicle (HEV). Many HEVs employ both an internal combustion engine and an electric motor. The goal is to provide lower emissions, while obtaining superior fuel economy and performance.; The objective of this thesis is to explore the use of game-theoretic principles in the implementation of supervisory control of a Hybrid Electric Vehicle (HEV). The work consists of two projects; the first deals with design of the controller itself, and the second, is modelling several test environments to test and redesign our controller in an iterative manner.; The simulation environment used is MATLAB/Simulink, a Mathworks product. Vehicle dynamics are modeled using CarSim (a Mechanical Simulation Corporation product). Real-time performance of the supervisory control algorithm is verified using RT-Lab (an Opal-RT product). One of the benchmark tools in REV research, ADVISOR (also MATLAB based), is referenced as a means of double-checking the consistency of simulation results using our customized test environment.; Simulation results using the Game-Theoretic Supervisory controller are compared with more conventional, benchmark approaches. Furthermore, a study of controller optimization is presented, as well as simulation results under many drive cycles. Additionally, HEV testing methods are investigated. Complex, comprehensive modeling is compared to simplified, analytical modeling of vehicle powertrains in terms of practical utility and accuracy.