Digital Implementation of Innovative Electric Drives

Nowadays high efficient electric drives are needed to achieve maximum performance of an electric motor. Moreover fault-tolerant capability of the drive has attracted great attention from research and industry world. Automotive sector for instance, is undergoing a radical shift from combustion engines to electric motors. Moreover, other sectors such as aircraft and navy, are moving towards the electrification of all the auxiliary services. An accurate motor control relies on the knowledge of the rotor position to achieve maximum performance of the electric drive. Sensors such as encoder or resolver are used to obtain the rotor position information, i.e. sensored control. In the last years, many algorithms have been developed and investigated to eliminate the position sensor which has some drawbacks. First of all the overall price is increased because additional hardware and cabling are needed. More importantly, the reliability of the drive is decreased since position sensor fault is common. In a sensorless drive, no position sensor is needed because the rotor position is estimated by processing stator currents or voltages according to the motor speed range. Sensorless control allows reducing the overall price, motor frame size and more importantly increasing the reliability of the drive. The latter, together with their fault-tolerant capability and high power density is increasing the attention on multiphase machines. This dissertation aims at investigating sensorless control at standstill and low speed by using high frequency injection techniques. Motor anisotropy is depicted by analyzing the current response due to the high frequency injection. Since the motor parameters are different, several responses are obtained from different motors. As a consequence, the performance achievable in sensorless control are different for each machine. The ability of the motor to be controlled in sensorless control is defined as self-sensing capability. The thesis can be divided into two main parts. In the first part, sensorless control applied to a multiphase machine with six phases is investigated. The machine is analyzed as two three-phase windings, each one supplied by a dedicated inverter. A fault-tolerant strategy is developed by studying the motor self-selfing capability. The novelty of the proposed research is that the self-sensing capability of the motor is enhanced and sensorless control without divergence is obtained. No complex compensation algorithms are applied and the fault tolerant capability is improved. Finally, an application of the Kalman fusion algorithm to the sensorless control of the machine is investigated. Sensor fusion is adopted to combine together more estimated positions from more winding sets. The purpose is to close all three-phase windings control loops on one fused position. By doing so, fault-tolerant of the machine is enhanced in the case of a fault in the position sensor. The second part of the thesis is focused on the developing of a novel observer transfer function for the rotating voltage signal injection. The novelty is that the whole estimator transfer function is retrieved in the Laplace domain by exploiting the modulation/demodulation theory. With this approach, the demodulation effects on filters transfer functions are considered and a proper observer transfer function is retrieved. As a consequence the observer regulator tuning allows achieving both maximum performance and the desired closed loop bandwidth. The whole thesis was fully validated through an intensive simulation and experimental stage, except the rotating voltage injection which was tested only by simulation. The aforementioned contents were presented by the author at several international conferences and IEEE Journal papers. The complete list of publications is reported at end of the dissertation.