Efficiency-Optimization Control of Extended Range Electric Vehicle Using Online Sequential Extreme Learning Machine

This paper describes the application of an Online Sequential Extreme Learning Machine(OS_ELM) for online efficiency-optimization control of Extended Range Electric Vehicle (EREV also called REEV). Efficiency-optimization control of EREV is formulated as a nonlinear constrained multi-objective problem with competing and non-commensurable objectives of fuel consumption, emissions, driving performance, battery life and driving range. To get real-time Pareto optimal solutions, an Offline Extreme Learning Machine and OS_ELM are hanged together. ELM is used to describe nonlinear system of EREV. When work status of gasoline engine or load change, optimum work status can be sought out by OS_ELM. Finally, the optimization is performed over the following three typical driving cycles that are currently used in the U.S. and European communities: 1) the Federal Test Procedure (FTP); 2) Extra Urban Driving Cycle (EUDC); and 3) Urban Dynamometer Driving Schedule (UDDS). The results demonstrate the capability of the proposed approach to generate well optimal solutions of the on-board charger optimization of EREV.