Adaptive Controller of Bidirectional DC-DC Converter Based on Extreme Learning Machine for Electric Vehicle

Abstract

This paper proposed a novel adaptive controller of a non-isolated bidirectional coupled inductor DC converter based on an extreme learning machine (ELM) for electric vehicles. MATLAB and SIMULINK are used as the software platform to simulate the proposed method. Different operating conditions are considered to show the efficacy of the extreme learning machine. From the simulation results, it is found that by designing the controller based on an extreme learning machine the controller can automatically adjust the value depending on the operating conditions (motor moving forward, backward, and regenerative braking). In addition, the performance of the motor with DC converter based on ELM is compared with DC converter based on grey wolf optimization and based on conventional PI controller. This is indicated by the overshoot and the settling time is much better when using controller based on an extreme learning machine (overshoot 221.6 and settling time 3.8 second).