Neural network approach to gain scheduling for traction control of electrical vehicles

This paper proposes a gain scheduling approach by neural network to force control of the electric vehicle wheels. To approximate to the reality in simulation, we utilize the traction force database of the motor, called the current-RPM-torque database, instead of the slip ratio measurements. The system is nonlinear and a constant gain cannot overcome all road conditions of the traction force control for the electric vehicles. The appropriate gains for different road conditions can be the training data of the neural network. In this paper, the proper parameters for the RBF neural network are obtained. The appropriate gains which have to fit the assigned specifications in time domain seem to be inverse proportion to the slip ratio slope.