Sensorless control of electrical drives

Abstract

ABSTRACT This paper provides an overview of advanced methods to perform sensorless control for D.C. and A.C. motors in detail. The features of the methods are presented, analyzed and evaluated. The conclusion can be used as a guideline for the persons interested in sensorless control. Keywords: Sensorless control, indirect measurement, electrical drives 1. INTRODUCTION As well known, the traditional method of a closed-loop motor control has to use one or more sensors to provide feedback, which makes the motor having a complicated structure, influences the operation reliability and increases manufacturing costs. In many hostile environments, position sensors cannot even be mounted. For small motors, the drawbacks become rather obvious, e.g. the sensors require additional wiring, which can also be vulnerable to electromagnetic interference to affect the performance. Such a sensor system rest rains the development of the motor at high speeds and at small sizes. As a result, a sensorless control is appeared and developed, namely, removing the sensor from the motor, applying some algorithms in the drive to monitor one or more key motor electrical parameters and to convert those data into position information, i.e. to get the position by indirect measurement. The sensorless control is robust against the interference and can enhance reliability, reduce manufacturing costs of the motor etc. Nowadays, the sensorless control is receiving wi de attention in the world [1]. There are two main categories of estimation methods for the sensorless control: one is based on the electromagnetic relationship and the other is based on observer techniques[2]. The first methods estimate the motor speed and rotor position by using the motor current and voltage signals detected. In these methods, the rotor position can be estimated by calculating the spatial vector of stator flux linkage, the terminal voltage and the phase inductance changing with the rotor position. Th e advantages of these methods are less calculation, simple and can be implemented easily, but the precision will decline at low speed, and they rely upon the motor parameters on a great scale. When motor parameters change due to temperature variation or magnetism saturation and so on, the precision and robustness will decline. The usual position-identification methods based on observer techniques include full order Luenberger observer, adaptive observer, variable structure observer, Kalman Filter etc[3].These methods have good dynamic performance, high stability and strong robustness. Th e shortcoming of these methods is that the eff ect of the speed adjustment is not ideal at low speed; what’s more, the algorithms are complex and require large amounts of calculating. But with the rapid development of high speed microprocessors in recent years, these methods which were considered impossible before have been carried out successfully.