The fuel economy of hybrid electric vehicles is very closely associated with the energy management strategy (EMS). In this paper, a practicality-oriented adaptive EMS for a parallel hybrid electric bus is presented, which combines the adaptive neuro-fuzzy inference system (ANFIS) and equivalent consumption minimization strategy (ECMS). Considering the regular and fixed route of the city bus, the optimal control trajectories can be attained by the dynamic programming in advance. Using the rolling optimization method, a group of optimal equivalent factors is extracted from aforementioned control trajectories and used as the training samples. Then, a trained ANFIS model that produces the optimal equivalent factor online is constructed, showcasing striking superiority in self-learning and inference. By applying the derived equivalent factor in the framework of the ECMS, an adaptive energy management controller is available to achieve desirable power distribution online. Finally, the simulation and hardware in the loop (HIL) tests are used to validate effectiveness and feasibility of the controller. The results demonstrate that, compared with other strategies, the fuel economy with the proposed strategy can be effectively improved.

Mode transition plays an important role in the operation of hybrid electric vehicles. The clutch is a key actuator for mode transition, but friction resulting from its use makes it difficult to achieve seamless mode transition. Moreover, as is usual in an electromechanical coupling system, the hybrid powertrain may exhibit complex dynamic behaviours as operating parameters change during mode transition. This paper considered changes in the clutch friction coefficient, and established a mathematical model of the hybrid powertrain in a parallel-series hybrid electric vehicle (PSHEV) during mode transition. The Routh-Hurwitz criterion and bifurcation theory are introduced to analyse the influence of speed difference on the dynamic characteristics of the hybrid powertrain, and the corresponding instability threshold were given. Furthermore, an optimised adaptive control methodology was proposed to control the dynamic behaviours, and tests on PSHEV were carried out. The results show that the variations of speed difference may lead to the occurrence of Fold and Hopf bifurcations, and that once the speed difference reaches the instability threshold, it may cause the system to enter a chaotic state and exhibit random oscillation. The proposed control methodology can effectively expand the security domain and suppress the oscillation to improve smoothness of mode transition.