Flux estimation of induction machines with the linear parameter-varying system identification method

In indirect field orientation control (FOC) methods, the magnitude and direction of the rotor flux are estimated from measurements of stator voltages, stator currents and the angular velocity of the shaft using a parameter model of the induction machine. However the performance of indirect FOC methods is dependent on the accuracy of the machine model and is therefore sensitive to variations in motor parameters such as the rotor resistance and the magnetizing inductance. Motor parameters vary greatly with temperature, frequency and current amplitude. This paper presents a novel method for estimating the rotor flux in an induction motor. Subspace identification methods are used to construct a linear parameter-varying (LPV), discrete time model of an induction motor based on measurements of the stator voltages and currents and of the angular velocity of the shaft. The identification algorithm has been tested on data obtained from a nonlinear, continuous-time simulation model.