Real-time implementation of sensorless indirect field-oriented control of three-phase induction motor using a Kalman smoothing-based observer

This paper proposes an indirect vector control scheme of an induction motor, using a Kalman smoothing-based observer to estimate speed. The observer is used to estimate the stator currents, rotor currents, and rotor mechanical speed. Observers based on conventional extended Kalman filters (CEKFs) depend on past output measurements of a system to predict its state variables at the next instant. In a smoothing-based observer, some future output measurements are also used to obtain a smoothed estimate of a past instant. The smoothed estimate, thus obtained, is used to predict and correct the states of the next instant. The performance of a CEKF-based observer largely depends on the proper determination of its measurement and process error covariance matrices. A trial and error method is usually engaged to arrive at these matrices. Smoothing helps to obtain a better state estimate compared with CEKF, with the same covariance matrices used in CEKF, which is obtained by trial and error. The improvement in estimation is mainly in the transient region. Estimates of low and zero speeds also show good improvement over those of CEKF. This betterment accomplished is without much increase in computational load. Experiments are conducted to compare the performance of a speed sensorless indirect vector control system with the observer, based on CEKF and smoothing for the same values of noise covariance matrices. The experiments conducted used various reference speeds, including low and zero speeds. Results show the superiority of the smoothing-based Kalman observer over CEKF-based observers.