Back-to-back competitive learning mechanism for fuzzy logic based supervisory control system of hybrid electric vehicles

This paper proposes a novel back-to-back competitive learning mechanism (BCLM) for a fuzzy logic (FL) supervisory control system for hybrid electric vehicles (HEVs). This mechanism allows continuous competition between two fuzzy logic controllers during real-world driving. The leading controller will have the regulatory function of the supervisory control system. Firstly, the configuration of the HEV model and its FL-based control system are analysed. Secondly, the algorithm of chaos-enhanced accelerated particle swarm optimization (CAPSO) is developed for back-to-back learning of the membership function. Thirdly, based on fuel-prioritized cost functions, the regulation of competitive assessment is designed to select a controller with a better fuel economy. Finally, the competitive performance of using the CAPSO algorithm is contrasted with other swarm-based methods and the BCLM-driven control system is validated by a hardware-in-the-loop test. The results demonstrate that the BCLM control system significantly reduces fuel consumption, at least 9% from charge sustaining and charge depleting based, and at least 7% from conventional FL-based systems.