Optimal braking control for an independent four wheel-motor-driven electric vehicle

Abstract

The lightweight, integrated, and high performance motor-wheel driven electric vehicle is clean, energy saving, and safe, and has the potential to form the ideal electric vehicle for the future. This paper proposes an optimal fuzzy neural network braking control strategy to determine the allocation of the front and rear regenerative braking torque and friction braking torque for the independent four Wheel-Motor-Driven (4WMD) electric vehicle. The proposed fuzzy neural network controller applies a five-layered neural network to map relations between the inputs (required total braking torque and battery State Of Charge (SOC)) and the outputs (front and rear in-wheel-motor regenerative braking torques), and uses a genetic algorithm to optimize offline the weights and thresholds of this fuzzy neural network to determine the dynamic allocation of the front and rear in-wheel-motor regenerative braking torque and friction braking torque. The objective is to maximize the battery SOC at the end of braking, and the constraints are the required braking speed of the vehicle, the limited regenerative braking torque of an in-wheel-motor, and the allowable SOC range of the battery. A lightweight independent 4WMD electric vehicle model (including vehicle longitudinal dynamic model, tyres, in-wheel motors, batteries and disc brakes) and this fuzzy neural network braking control model are built in the Matlab/Simulink environment, and are simulated and validated under different braking scenarios. The simulation results illustrate that this optimal braking control is better than the simple rule-based braking control, recovering more regenerative braking energy while meeting the vehicle braking performance requirements.