GA-based neural network for energy recovery system of the electric motorcycle

• Published in: Expert systems with applications 2011-04, Vol.38 (4), p.3034-3039
• Main Authors: Cheng, Chin-Hsing, Ye, Jian-Xun
• Format: Article
• Language: English
• Subjects: Batteries; Computer simulation; Control systems; Electric batteries; Electric motorcycles; Energy recovery; Genetic algorithms; Matlab; Neural network; Neural networks; Regenerative; Regenerative braking
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2010.08.093

► Increases the traveling distance of the vehicle. ► Improves the performance and life-cycle of batteries. ► The energy recovery of batteries becomes more stable. This paper discusses a regenerative braking system for the electric motorcycle that performs regenerative energy recovery based on neural network control with a boost converter. A constant regenerative current control scheme is proposed, thereby providing improved performance and high energy recovery efficiency at minimum cost. The neural network controller is used to simulate the regenerative system in Matlab/Simulink and neural network toolbox. We can sieve out the suitable training samples to obtain good performance of the controllers, and the neural network with genetic algorithms is used to design the controller. Simulation results of neural network controller show a more steady quality and extended time of charging. The proposed scheme not only increases the traveling distance of the vehicle but also improves the performance and life-cycle of batteries, and the energy recovery of batteries becomes more stable. Therefore, the market of the electric vehicle will become more competitively.