High performance rotor position sensorless control of switched reluctance machines over a wide speed range

Switched reluctance machines are doubly salient machines. However, successful operation of switched reluctance machines depends heavily on the knowledge of accurate rotor position. The rotor position can be obtained either directly from position sensors or indirectly from position estimation methods. Direct rotor position sensors increase both the cost and size of the overall drive system, and may bring reliability problems in some operating environments. Indirect rotor position estimation is an attractive alternative.; In this dissertation, a fuzzy generalized nonlinear magnetic model is first proposed. Combining two magnetizing curves at aligned and unaligned rotor positions, the generalized nonlinear magnetic model can develop detailed magnetizing curves for most switched reluctance machines. Based on detailed magnetizing information, two two-level rotor position estimation algorithms are presented. The two levels reinforce each other, and the result is a very accurate estimated rotor position. The proposed rotor position estimation method performs very reliably over a wide speed range from zero to high speeds ({dollar}>{dollar}2000 rpm).; Finally, a high performance sensorless variable-speed controller for switched reluctance machines is suggested. The variable-speed controller consists of a modified PI current regulator and a feedforward turn-on angle controller. The modified PI current regulator can generate different command phase currents according to different quadrant operations. The feedforward turn-on angle controller generates the command turn-on angle based on the actual and command rotor speeds. Simulation and experimental results show that the proposed sensorless variable-speed controller has very good dynamic and steady state performance, and the proposed controller also meets the requirements for high efficiency operation of switched reluctance machines.; The main contributions of this research are: (1) A fuzzy generalized nonlinear magnetic model is proposed, which saves significant effort and time in control algorithm development and evaluation. (2) The proposed two-step rotor position estimation algorithms can accurately estimate the rotor position, which makes the high performance position sensorless control of switched reluctance machines be possible. (3) The novel position sensorless variable speed controller greatly enhances the speed regulation performance, which increases the competition capability of switched reluctance machines in high performance machine drive arena. (Abstract shortened by UMI.)...