In More Electric Aircraft applications, Dual Three-Phase Permanent Magnet Synchronous Machines can potentially be employed in starter/generator systems due to their high-power density and fault tolerance capabilities. However, achieving robust sensorless control and fault-tolerant operation in these machines remains a significant challenge, particularly under open-circuit fault conditions and parameter variations. This thesis addresses these challenges by introducing a novel Model Reference Adaptive System sensorless control strategy and an adaptive Proportional-Integral-Resonant fault-tolerant control scheme, ensuring high reliability and superior performance in More Electric Aircraft applications. The proposed Model Reference Adaptive System sensorless control strategy effectively estimates rotor position and speed, replacing conventional mechanical sensors which may not be viable in some specific aerospace environments A Forgetting Factor Recursive Least Squares online parameter estimation method is integrated into the Model Reference Adaptive System framework, significantly enhancing estimation accuracy while reducing computational complexity. Unlike traditional recursive least squares approaches, the Forgetting Factor Recursive Least Squares algorithm eliminates matrix inversion computations, making it computationally efficient and well-suited for real-time applications. Additionally, a Proportional-Integral-Resonant-based fault-tolerant control strategy is developed to mitigate electromagnetic torque ripple under open-circuit fault conditions. By modifying the current reference in the faulted phase and incorporating resonant compensation, the proposed control strategy effectively suppresses second-harmonic current distortions, ensuring smooth operation despite fault conditions. The proposed control methodologies are validated through comprehensive simulations and real-time experimental studies using a laboratory-based Dual Three-Phase Permanent Magnet Synchronous Machine prototype. When a single open-circuit fault is present, the torque of the dual three-phase machine studied in this work is derated to approximately 57.6% of the rated value. Under such a single open-circuit fault with realistic per-phase current limiting, simulations show that the proposed Proportional-Integral-Resonant-based fault-tolerant controller reduces the worst-peak torque error by about a factor of two compared with a standard Proportional-Integral controller, and a similar improvement is observed experimentally at 900 rpm. These results indicate that, relative to a conventional Proportional-Integral scheme, the Proportional-Integral-Resonant-based method provides stronger torque-ripple suppression under practical single open-circuit fault conditions while maintaining stable sensorless operation. For the Model Reference Adaptive System estimator, the Forgetting Factor Recursive Least Squares-enhanced Model Reference Adaptive System converges within 1 ms in simulation, keeps the identified parameters within 0.1% of their nominal values, and removes the approximate 2° position bias that appears with a fixed-parameter Model Reference Adaptive System using conventional Recursive Least Squares. Experimental results further confirm that, after step changes in the d- or q-axis current commands, the estimated rotor flux and stator inductance quickly reconverge to their nominal values with only small rotor-position errors. Compared with a traditional Recursive Least Squares implementation, the proposed Forgetting Factor Recursive Least Squares algorithm therefore improves robustness to parameter variations while avoiding matrix inversion. The results confirm that the proposed Model Reference Adaptive System sensorless estimation for Proportional-Integral-Resonant-based fault-tolerant control and Forgetting Factor Recursive Least Squares online parameter estimation strategies provide enhanced system robustness, improved estimation accuracy, and superior torque performance in More Electric Aircraft applications.
Shengyu Cao (Wed,) studied this question.