Driver-oriented intelligent control methodology for series-parallel hybrid electric vehicles

Rapid development in informatics enables optimization of hybrid electric vehicle (HEV) systems with a fusion of external dynamics, e.g. driver and traffic. This thesis studies the driver-oriented energy management problem of a series-parallel HEV for promoting a paradigm shift to more sustainable mobility. The objective of this research is to characterize human driving behaviour and maximize personalized energy economy for real-world driving. Driver-oriented intelligent control methodology is proposed to tailor a family of control strategies for minimizing energy consumption of HEVs. The research uses emerging ‘mediums’ computational intelligence and Internet of the Things to create an accessible human-machine interaction system for energy management personalization. The research work is carried out in four parts with distinctive contributions. Firstly, a novel approach that uses personalized non-stationary inference is proposed to increase the robustness of the rule-based vehicle control system through real-time driving behaviour monitoring for vehicle energy economy improvement. On the basis of the personalized non-stationary inference, the author aims to transfer driving style classification methods from continuous indexing towards discrete classes and expand the human-related factors from velocity and acceleration only towards velocity, gas pedal, brake pedal, and steering wheel angle. Secondly, the concept of the driver-identified supervisory control system is introduced, which forms a novel architecture of adaptive energy management for HEVs. As a man-machine system, the proposed system can accurately identify the human driver from natural operating signals and provides driver-identified globally optimal control policies as opposed to mere control actions. To better acclimate to stochastic driving condition, the author considers elevating HEV energy management into an online level. Thirdly, a novel back-to-back competitive learning mechanism is proposed for a fuzzy logic supervisory control system for HEVs. This mechanism allows continuous competition between two fuzzy logic controllers during real-world driving. Not only the optimizer, but also the predictor is considered to promote for synchronous online update. Fourthly, an online predictive control strategy for series-parallel plug-in HEVs is investigated, resulting in a novel online optimization methodology named the dual-loop online intelligent programming that is proposed for velocity prediction and energy-flow control. All work is demonstrated via customized experimental plans which are designed based on a hardware-in-the-loop testing bench and a driving simulator platform. This allows a deeper insight into each control strategy in the driver-oriented intelligent control methodology, exposing strengths and drawbacks that have not been noticeable from past work