Institute for Drive Systems and Power Electronics Research Research Projects
Self-Sensing Control of Permanent Magnet Synchronous Machines for the Complete Speed Range

Self-Sensing Control of Permanent Magnet Synchronous Machines for the Complete Speed Range

Led by:  Prof. Dr.-Ing. Axel Mertens
Team:  Dr.-Ing. Karsten Wiedmann
Year:  2012

For several years already, the IAL is doing research work on sensorless control of permanent magnet synchronous machines (PMSM). In this respect, an anisotropy-based estimation of the rotor position is carried out for the lower speed range. In other words, the dependency of the phase inductances on the rotor position is evaluated. For higher speeds, the estimation of the rotor position is based on EMF evaluation.

Sensorless control of PMSMs has been a main research focus in the field of drive control for more than 20 years. Numerous publications have dealt with basic methods like HF injection, INFORM method, etc. and the mitigation of possible distortions. Those topics that still require more research work to be done, are specified in the following:

1. In general, anisotropy-based and EMF-based estimations of the rotor position rely on different structures, thus often complica­ting the combination of different methods and increasing the computational effort.

2. For anisotropy-based methods, it is important to minimise the influence of a dynamic current control, so that the dynamics of the superior torque and speed control need not be reduced. This generally requires the know­ledge of the saturation-dependent inductance and/or elaborate filter structures.

3. Sensorless system identification is of great importance in order to consider or minimise the influence of saturation effects (see item 2), multiple saliencies or nonlinearities of the converter.

4. The computational effort of the methods in use, shall be kept to a minimum. But many of the proposed methods include divisions by variables or matrix inversions.

Considering all of these aspects in one method is quite difficult. This is one reason why sensorless control in PMSMs has not become as accepted in industrial applications as would actually be possible. In general, a good knowledge of the drive system is required, in order to obtain sensorless control of sufficient quality.

As part of a PhD dissertation, a method was developed at IAL which is much more advantageous with respect to the criteria mentioned above, compared to the state-of-the-art. The method is based on the Model Reference Adaptive System (MRAS) approach, where a flux observer monitors the stator flux linkage. Based on the observer’s estimation error, the rotor position for the complete speed range as well as crucial system parameters can be identified. For this purpose, the gradient descent method is used which is known for its low computational effort. This self-sensing MRAS (SS-MRAS) approach is of modular design. Particular features, like for example the identification of transient inductances in the operating point or the reduction of harmonic errors of the position estimation, can be considered additionally, if the drive system is not sufficiently well known. In this respect, an autocommissioning was developed for motors at zero speed, where no specific operating points have to be assumed.