Kriging assisted on-line torque calculation for brushless DC motors used in electric vehicles

Torque is one of the most important control factors for a vehicle´s motion. Compared with internal combustion engine, electric motors can have a more accurate torque feedback. In electric vehicles, direct torque control of the permanent magnet synchronous motor has been studied. However, because of non-ideal back EMF phenomenon, direct torque control of brushless DC motors has not been widely used. In this paper, a new method using kriging to calculate back EMF in a real-time on-line fashion is presented. Kriging prediction is used to approximate the back EMF of the motor based on information from sampled points. With motor speed and rotor position as inputs, kriging predicts back EMF as the output that is used to calculate the motor torque with three phase currents. Using this novel method, motor torque can be accurately calculated and implement in micro control units of vehicles, even when facing extremely high/low temperature and aging conditions.