Research on multi objective intelligent shifting schedule of electric vehicle AMT considering ride comfort and economy

Abstract

Considering the current shifting strategy of multi-speed automatic manual transmission (AMT) separates the steady-state shifting from the transient shifting process in the pure electric vehicle, it is difficult to find a comprehensive improvement of shifting quality, dynamic performance, and driving economy. In this paper, taking advantage of the artificial intelligence technology, a fuzzy neural network (FNN) based T-S model is established via obtaining the training data from skilled drivers' experience and expert knowledge. A two-speed AMT pure electric vehicle model is used to investigate the fuzzy shifting strategy performance. According to the co-simulation results of AMESim and SIMULINK, the average jerk of 10.006 is recorded, compared to the value of 16.472 based on an ordinary shifting schedule. The results show that FNN-based schedule fully reflects drivers' shifting intentions in pursuing shifting smoothness, at the same time, improving vehicle dynamic performance with negligible economic performance loss.