From Off-line to On-line Control of a Multimode Power Split Hybrid Electric Vehicle Powertrain

Abstract

Abstract On-line optimal control represents a crucial issue in the development of multimode power split hybrid electric vehicles (HEVs). Finding a control strategy that guarantees fuel economy optimality and ease of implementation still reveals an open research question. This paper aims at developing an on-line control approach for multimode HEVs based on previously implemented off-line control. The two control levels for multimode HEVs are presented: the operating mode selection and the torque split determination. The former is addressed adopting a machine learning approach where artificial neural networks (NNs) are trained in supervised learning using off-line control data. The torque split is resolved on-line according to efficiency-based maps extracted off-line. Simulation results for a specific multimode HEV design demonstrate the effectiveness of the developed control strategy in minimizing the value of predicted fuel consumption. Furthermore, a sensitivity study is conducted for the NN sizing parameters. The ease of implementation and adaptability suggests the potential application of the developed online control approach in a design methodology for multimode HEVs.