Wide Speed Range Sensorless AC Motor Drives, including Operation at Zero-Frequency, without using Test Signals

A sensorless electric motor drive is the popular term for drives which do not use shaft mounted speed or position sensors. Sensorless operation is highly desirable for reasons of cost, simplicity and system integrity. However, it is well known that there are serious problems with sensorless motor drive control at zero and low speeds and this has been one of the main research topics in this field for many years. The conventional method for sensorless control, used in commercial products, is to estimate the machine flux and speed using a mathematical model of the motor. Below 1 to 2% base speed however, position and speed estimation using such a model deteriorates and speed and torque control is lost. There has been a recent impetus for zero speed sensorless drives for more-electric aircraft and vehicular applications. For the former, there is a requirement for direct electromechanical (EM) actuation of critical actuators in which locking of the mechanical transmission is not permissible. In the vehicular field direct EM drives will be required for the main drive train, and for power steering, active suspension and braking actuation. One approach to the solution of the zero speed problem, which does not require a machine model, has been to exploit the natural asymmetries or saliencies in AC machines. These saliencies are cause by magnetic flux saturation and the geometry of the construction of the motor itself. Flux or rotor position can then be tracked by processing the current response to a test voltage signal injection overlaid on the supplied motor voltage. These signal injection methods are now quite well understood, but do contribute to increased accoustic noise, reduced efficiency, the requirement for additional sensors, and an increase in bearing wear and electrical stress within the machine windings.The current proposal aims to overcome the above disadvantages by developing methodologies by which:1) No signal injection is required, the method being integrated with the fundamental voltage applied to the drive via the power converter. This eliminates the problems of extra noise, losses, bearing wear and electrical stresses.2) The requirements for sensors is substantially reduced (depending on the application). For bespoke applications (e.g. aerospace, automotive), the aim will be for one current sensor and one low cost di/dt sensor. For industrial standard drives the target aim is to use only the existing line current sensors. These aims are quite challenging. Mathematical feasibility of a non-signal injection method has been shown at Nottingham and the technique is currently subject to patent at the University. Practical investigation is now possible owing to advances in high-accuracy timing and sampling available in low-cost digital control systems.