A Q-learning Approach for SoftECU Design in Hybrid Electric Vehicles

A software electronic control unit (SoftECU) implements simplified versions of the algorithms coded in a corresponding real ECU. In this paper a Q-learning approach for the design of SoftECUs is proposed. The training phase of the SoftECU is based on the mismatch between the commands provided by the real ECU and those coming from the SoftECU model. The technique is applied to the case of an energy management ECU for hybrid electric vehicles. The effectiveness of the SoftECU is validated by considering the new European driving cycle and the worldwide harmonized light vehicles test procedure.