Improved flux and torque estimators for application in direct torque controlled IPM synchronous motors at very low speed

摘要

A number of essential and important improvements of a direct torque controlled interior permanent magnet (IPM) synchronous motor drive are presented in this thesis. The studies comprising of analytical, modelling and experimental implementation indicate clearly the potential of a high-performance direct torque controlled IPM synchronous motor drive without any rotor position or speed sensor.The direct torque control (DTC) technique, its associated problems and algorithms to overcome these problems have been investigated in this thesis. Analysis and compensation of non-linear effects of the inverter such as forward voltage drop and dead-time, speed sensorless control and extraction of accurate rotor position information for reliable sensorless operation of a direct torque controlled IPM synchronous motor drive, especially at very low speeds, are of major concern in this thesis.Two different speed estimation algorithms, a rotor flux linkage vector and an adaptive sliding mode based observers, were implemented on space vector modulated (SVM) DTC IPM drives. Experimental results show good performance of the observers but their performance at very low speed becomes unsatisfactory, mainly due to non-linearities of the inverter.Compensation schemes for the non-linear characteristics of the inverter were proposed. Experimental results show significant reduction in stator flux linkage and torque estimation error, leading to improved low speed performance of the DTC IPM drive. However, extremely low speed including zero speed sensorless operation of the DTC IPM drive is not achievable using the offline compensation technique.A high frequency signal injection algorithm is proposed, where a carrier excitation signal is superimposed on the d-component of the stator voltage in the estimated rotor reference frame and an alternating high frequency current response is detected in the q- direction of the same frame with its amplitude modulated by the rotor position estimation error. The stator flux linkages and torque are estimated using the current model of the IPM machine with the obtained rotor position information. Modelling and experimental results show that rotor position and speed can be estimated very accurately at very low speeds including standstill using the proposed algorithm. A changeover algorithm, which uses a weighting function, is proposed and implemented for smooth handover of observers during transitions between low and high speeds.The abovementioned studies show that the DTC techniques for IPM machines have now been further improved and are closer to being a viable and cost-effective candidate for sensorless PM synchronous motor drives.