CARRIER SIGNAL INJECTION BASED SENSORLESS CONTROL OF PERMANENT MAGNET BRUSHLESS AC MACHINES

This thesis is focused on the carrier signal injection based sensorless control of Permanent Magnet (PM) Brushless AC (BLAC) machines. Based on the machine saliency property, carrier signal injection based sensorless techniques have been well developed in the past decade. In order to provide the insight into machine saliency information, a simplified experimental procedure is presented to obtain the machine saliency distribution in the dq plane, including magnetic saturation and cross-saturation effects. Based on the measured machine saliency information, Sensorless Safety Operation Area (SSOA), which accounts for quantization errors in the Analog to Digital (AD) conversion, is proposed to investigate the effectiveness of sensorless operation for practical applications. The SSOA defines a working area in dq plane, in which the machine can work in sensorless mode with guaranteed steady state performance for either pulsating or rotating carrier signal injection based methods. The non-ideal characteristics of machines and drives introduce additional carrier current disturbances in carrier signal injection based sensorless control, which eventually results in the deterioration of the position estimation accuracy, the degradation of dynamic performance and even stability problems. This thesis investigates the influence of inverter nonlinearity effects on rotating carrier signal injection based methods. With the aid of theoretical analysis and experimental measurement, it is proven that the positive sequence carrier current distortion resulting from inverter nonlinearity effects can be used to compensate the influence of inverter nonlinearity on negative sequence carrier current. Hence, a new online post-compensation scheme is developed in this thesis by utilizing the measured distortion of positive sequence carrier current. In sensorless control, great efforts are required to compensate for nonlinear effects in order to improve the accuracy of the estimated position information. Alternatively, the optimal efficiency of machines, instead of accurate position estimation, can be taken as the sensorless operation objective. In this way, the sensorless operation performance is improved by the proposed online optimal efficiency tracking without any compensation for nonlinear effects. In addition, a robust magnetic polarity identification scheme is developed in this thesis based on d-axis magnetic saturation effect. The proposed scheme can be seamlessly integrated into conventional carrier signal injection based sensorless control algorithm.