Reformulated Predictive Torque and Flux Control With a Full-order Adaptive Observer and Accurate Discrete-time Models for Sensorless Induction Machine Drives

Abstract

In this paper, we present a reformulation of both the predictive torque and flux control (PTC) scheme and the full-order adaptive observer (FAO) for induction machine drives. The proposed approach is based on a state-space representation expressed exclusively in terms of stator current and stator flux linkage, simplifying the observer structure and removing the explicit dependence on rotor flux variables found in conventional sensorless formulations. This representation is consistently applied within both the FAO and PTC frameworks, and second- and higher-order discrete-time models are derived using Taylor- and Runge-Kutta-based methods to enhance numerical accuracy and dynamic performance. The resulting FAO-PTC scheme is validated through Hardware-in-the-Loop simulations, demonstrating steady-state performance comparable to conventional designs, faster transient response, improved dynamic behaviour, and a reduced state-space order, albeit with slightly higher computational cost. Notably, simply employing a more accurate observer substantially enhances the performance of the sensorless scheme. Among the evaluated discretization strategies, the Taylor-based model provides the highest steady-state accuracy and fastest convergence, with only a modest increase in torque ripple. Overall, the proposed reformulated FAO-PTC framework achieves a balanced trade-off between accuracy, implementation simplicity, and computational efficiency for real-time sensorless induction machine drives.