Dynamic control of V-belt continuously variable transmission-driven electric scooter using hybrid modified recurrent legendre neural network control system

Abstract

Because of unknown nonlinear and time-varying characteristics of V-belt continuously variable transmission (CVT)-driven electric scooter by using permanent magnet synchronous motor (PMSM) servo drive system, all gains tuning process for linear controller is a very time-consuming task. A hybrid modified recurrent Legendre neural network (NN) control system, which consists of an inspector control, a hybrid modified recurrent Legendre NN control and a recouped control with estimation law, is proposed for controlling the V-belt CVT-driven electric scooter under the occurrence of the nonlinear load disturbances and the variation of parameters to acquire better control performance. Moreover, the online parameters tuning method of the modified recurrent Legendre NN is based on Lyapunov stability theorem and gradient descent method. Furthermore, the two optimal learning rates of the hybrid modified recurrent Legendre NN control system are derived according to discrete Lyapunov function to enhance convergence speed. The proposed control scheme is capable of responding to system’s nonlinear and time-varying behaviors due to online learning ability. Finally, some experimental results are verified to show that the effectiveness of the proposed hybrid modified recurrent Legendre NN control system controlled the V-belt CVT-driven electric scooter by using PMSM servo drive system.