Hybrid Optimization For New Energy Automobile Power Train

This study provides an optimization methodology in powertrain configurations and energy management strategy for a multi-mode plug-in hybrid electric vehicle.The challenge in this study is that the energy management strategy is highly intertwined with the powertrain configurations and all the optimization depends on the velocity profile which is not viable due to the uncertainty of traffic.The problem-specific complexity of this powertrain makes it even more tricky.Based on the analysis of the powertrain structure,a simulation model is developed which is then validated via commercial software-AVL Cruise.Two self-adaptive solutions in the proposed model,a fuzzy PID driver and a self-adaptive shifting schedule are established to accommodate the powertrain parameter variations during the particle swarm optimization(PSO)process.For a powertrain in this degree of complexity,multiple rule-based energy management strategy(EMS)can be established.To select an EMS that is competent with the later proposed hybrid optimization algorithm and to consider the driving pattern and enhance diving comfort,a novel third objective is proposed then exploited to optimize two most likely EMSs.In order to obtain optimal and applicable results of both the powertrain and the energy management strategy configurations,a novel hybrid optimization algorithm is proposed which optimizes the powertrain configurations via PSO while,for the inner optimizer,a two-layer optimization controller is proposed.In the first layer,sub-optimal results are obtained from ant colony optimization in continuous domain(ACOR).Additionally,the second layer is designed to final-confirm both mode changes and gear shifting.The efficacy of the proposed hybrid optimization algorithm is then validated via two possible applications: parallel hybrid electric vehicle and an all-electric vehicle which present two different complicity of optimization.It is concluded that the proposed algorithm is most effective for complex powertrains,especially for the studied multimode hybrid electric vehicle.At last,two possible applications—modifications to the shifting schedule and energy management strategy,or furthermore,developing a neuro network controller utilizing the optimal results are analyzed.To remedy the velocity prediction outcome,a hybrid prediction method using electromyogram(EMG) signals is then proposed.The results suggest that the hybrid optimization presented in this thesis provides a solution to squeeze the potential of a PHEV while maintaining its practicability.