Switched reluctance motors (SRMs) are promising candidates for electric vehicles due to lower manufacturing costs, higher efficiency, and robustness operation in a harsh envi-ronment. For accurate control of the SRM, the real-time rotor position is needed for phase computation. To obtain position information, position-sensorless control techniques have been developed to take the role of position sensors in commercial SRM drives for cost reduction or sensor-fault tolerance capability. Nowadays, the position-sensorless control of SRMs still suffers from a technical problem: the dependence on magnetic characteris-tics. Existing position estimation algorithms often require time-consuming offline meas-urement of magnetic parameters, limiting the broad applications due to the low generality. It is therefore of great significance to develop universal position-sensorless control tech-niques with less magnetic parameter dependence.
Zero- and low-speed position-sensorless control of the SRM needs high-frequency in-jection into the idle phase to measure the stator inductance. Rotor position is often esti-mated from the prestored inductance lookup table but is replaced by a new regional phase-locked loop (RPLL) with a self-commissioning process in this thesis. The modeling of the unsaturated stator inductance can be established automatically via the pulse voltage injection at the initial stage without offline testing. The RPLL embedded with a three-phase heterodyne design can estimate the full-cycle rotor position from the idle-phase in-ductance based on the unsaturated inductance model. The proposed low-speed position estimator can also realize robust sensorless control in four-quadrant operation and magnet-ic saturation conditions without complicated magnetic characteristics. Besides, local sta-bility of the position estimator is proved, and an optimized parameter design scheme is given.
Although pulse voltage injection offers accurate position estimation in low-speed op-eration, the induced pulse current results in additional copper loss and torque ripples. This problem is overcome in the thesis by regulating the magnitude of induced current at a minimal level. The induced current regulator is designed as a terminal sliding-mode con-troller that adjusts the injection voltage online over the whole idle-phase period. Proper control parameter selection based on the convergence analysis and stability proof ensures robust control performance against parameter uncertainties. The proposed pulse injection scheme combined with the RPLL can guarantee accurate position estimation while reduc-ing copper losses and torque ripples significantly.
Due to the shortened idle-phase duration when the rotor speed increases, pulse injec-tion methods are infeasible for high-speed position estimation. To solve the problem, this thesis proposes a nonlinear observer based on feature position estimation in conduction phases for high-speed sensorless control. A self-commissioning method is adopted to cap-ture a two-dimensional flux linkage curve at a feature position, which avoids offline measurement of the complete three-dimensional characteristics. However, the estimated feature position has low resolution, and its estimation accuracy is degraded by nonideal flux linkage errors. To improve the sensorless control performance, a nonlinear state ob-server using online Fourier series is then designed to eliminate disturbances in position es-timation. Parameter design based on a small-signal analysis is also given to guarantee ac-curate position and speed estimation.
High-speed position-sensorless control is further simplified using a new quadrature flux estimator without using any flux linkage characteristics. The method requires neither offline measurement nor online self-commissioning. This advantage is realized by adopt-ing a speed-adaptive bandpass filter to extract the fundamental flux linkage. A three-phase phase-locked loop is then used to estimate the rotor position from the orthogonal flux linkage signals without a priori knowledge of the SRM magnetic characteristics. The magnetic-parameter-free position estimation can facilitate the application of sensorless control in a general-purpose SRM converter.
A wide-speed range position estimation scheme is realized by combining both the low-speed and high-speed position estimation approaches. Consequently, a universal posi-tion-sensorless control scheme is proposed in the thesis, covering the full-speed range and not requiring offline measurement effort.
The proposed position estimation schemes are verified on a 5.5 kW 12/8 SRM test bench. / Thesis / Doctor of Philosophy (PhD)
| Identifer | oai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/26874 |
| Date | January 2021 |
| Creators | Xiao, Dianxun |
| Contributors | Emadi, Ali, Electrical and Computer Engineering |
| Source Sets | McMaster University |
| Language | English |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0021 seconds