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 engines, electric motors can have a more accurate torque feedback which brings a lot of advantages to vehicle dynamics and stability control. However, motors used in electric vehicles are facing more difficult conditions than those in conventional applications, such as extreme high/low temperature changing, vibration, aging, etc. The variation of motor parameters due to harsh working conditions can lead to serious problems for motor torque estimation and thus dynamic control of electric vehicles. In this paper, a new method using kriging to estimate the back EMF and thus accurately calculate motor torque in an on-line fashion is presented. 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 even facing high/low temperatures or aging conditions. Experimental tests under the high temperature have been conducted to verify the applicability of the proposed method.