Markus Seilmeier

Sensorless Control of PMSM for the Whole Speed Range Using Two-Degree-of-Freedom Current Control and HF Test Current Injection for Low-Speed Range

In this paper, an innovative sensorless two-degree-of-freedom current control scheme for the whole speed range is proposed. It consists of a model-based dynamic feedforward control to set the reference response and model reference tracking controllers providing disturbance rejection and the position error signals needed for sensorless control. For low- and zero-speed operation, alternating test current injection is used to gain a position error signal. A flatness-based test signal precontrol provides compensation for secondary saliencies like cross-saturation and higher harmonics. For mid- and high-speed operation, it is shown how the model-based dynamic feedforward control can be modified to obtain a high-quality position error signal from the tracking controller. No additional model-based estimator evaluating back-EMF information is needed. The effectiveness of the proposed method is proven by experimental results.

Flatness based sensorless control of PMSM using test current signal injection and compensation for differential cross-coupling inductances at standstill and low speed range

This paper presents a novel flatness based two-degree-of-freedom (2DoF) sensorless control scheme for PMSM at standstill and for low speed range using test current signal injection in the estimated d-axis of the field oriented dq-coordinate system. Asymmetrical winding and saturation might cause the occurrence of differential cross-coupling inductances and as a result the estimated rotor position deviates from the real value. The differential cross-coupling inductances have to be compensated for to overcome this problem. For that a flatness based dynamic feed-forward control is proposed. The control error due to model uncertainties is minimized using a controller that fulfills the internal model principle. The control signal of the high frequency controller in the q-axis is demodulated and afterwards used as the input signal of a tracking-observer to estimate the rotor position.

Identification of high frequency resistances and inductances for sensorless control of PMSM

Recently published papers show that also a saliency in the high frequency (HF) resistances can be used for sensorless control. Those papers assume a diagonal HF resistance matrix. The aim of this paper is to show, based on experimental results, that a phenomenon like cross-saturation for the inductances is also existent for the HF resistances. The voltage drops caused by the HF resistances are typically very small and thus their identification is challenging. Therefore a high precision 3-phase current measurement is used and the dead-time and voltage drop errors caused by the inverter are compensated for. A high frequency test current control allowing zero steady-state control error is needed to overcome the limitations of voltage injection based methods.

Modelling and model based compensation of non-ideal characteristics of two-level voltage source inverters for drive control application

In order to be able to optimize the performance of sophisticated model based AC machine control, precise models of the overall drive system are needed. This way compensation schemes for non-ideal characteristics can be derived based on the model inversion technique. Power electronic converters show the following non-ideal effects which have to be considered: turn-on delay time of the gate drive to prevent dc-link short-circuiting, non-ideal switching characteristics and forward voltages of the power electronic devices. In this paper models for these disturbances are derived and the respective compensation schemes are deduced. In constrast to conventional approaches for converter linearization, an analytic compensation law which allows considering different forward voltage-current characteristics of the diode and transistor is proposed. The effectiveness of this method is proven by means of experimental results.

PMSM model for sensorless control considering saturation induced secondary saliencies

Saliencies in electric machines are caused by the machine geometry and saturation. PMSM typically do not just show a single sinusoidal saliency, which is the main saliency utilized for position tracking. Secondary saliencies are a major source of deterioration for sensorless control, whereby saturation is reported to be the main problem. Cross-saturation, caused by a movement of the main flux away from the d-axis, results in steady-state position estimation errors. Moreover saturation can result in 6n-harmonic components in the inductances in field oriented coordinates. This effect causes position dependent estimation errors. Both effects need to be compensated for, if significant. For model based compensation a machine model with time-variant parameters is needed to cope with nonlinear material characteristics. A general machine model in field oriented coordinates considering multiple saliencies is derived and a method for systematic determination of the position dependent inductances is proposed. The machine model is verified by test bench measurement results.

Modeling of PMSM with multiple saliencies using a stator-oriented magnetic circuit approach

Precise modeling of PMSM needs considering the nonlinear characteristics of the hard and soft magnetic materials used in the machine. Therefore nonlinear machine models with time-variant parameters are used. Usually saturation is modeled in the field oriented coordinates assuming that the d-axis flux affects the q-axis flux and vice versa. If certain conditions are fulfilled cross saturation inductances can occur. This paper introduces a new modeling approach to consider multiple geometric and saturation induced saliencies based on a stator-oriented magnetic circuit approach. The advantage of this method is that all kind of saliencies that occur in the electrical machine can be considered simultaneously. Moreover modeling in the stator coordinates leads to a physically motivated explanation when the cross saturation effect can occur in a PMSM. The result of the new modeling approach are new insights for the interpretation of the resulting inductances in the field oriented model.

Model based closed loop control scheme for compensation of harmonic currents in PM-synchronous machines

With increasing use of concentrated windings in permanent magnet synchronous machines harmonic effects have to be considered more and more for current control design. A structural expansion of the control scheme is needed to handle these higher harmonics. In this paper an effective method is proposed to carry out an expansion of the control scheme. Firstly the nonlinear disturbances are transformed into new fictitious rotating coordinate systems to get an easier description of the disturbances. The compensation of the harmonics is achieved in the new coordinates by applying closed loop controllers with suitable disturbance models. Therefore the control scheme is robust concerning parameter variations. In order to decouple reference and disturbance response a nonlinear model based dynamic feed forward control is used. The proposed control scheme is proven by test bench measurements.

Identification of steady-state inductances of PMSM using polynomial representations of the flux surfaces

Sophisticated model based control strategies for permanent magnet synchronous machines (PMSM) require precise modeling and knowledge of the machine parameters to achieve a high performance drive control. Nonlinear material characteristics of the iron and permanent magnets used in the machine lead to time-variant modeling approaches to deal with those phenomena. Typically iron losses are not considered in control plant modeling. However, it has already been shown that iron losses and non-ideal characteristics of the power electronic converter can be a major source of deterioration for the steady-state PMSM parameter identification. In this paper an elegant two-step identification strategy is proposed which minimizes the impact of iron losses and compensates for the influence of converter voltage errors during the identification of the flux surfaces (first step). In the second step the steady-state inductances are identified using polynomial representations of the flux surfaces. Like this, in contrast to conventional methods, the inductances can be identified even for zero current.

Carrier signal based sensorless control of electrically excited synchronous machines at standstill and low speed using the rotor winding as a receiver

This paper presents a new carrier signal based approach for motion sensorless control of electrically excited synchronous machines. In the case of permanent magnet synchronous machines sensorless control at standstill and low speed is mostly done by injecting a high-frequency voltage and evaluating the resulting current response. Hence, the stator is used as the transmitter and as the receiver of the carrier signal. The major drawback is that a magnetic saliency is mandatory. In the case of electrically excited synchronous machines the rotor voltage serves as an additional input to the system and the resulting field current as an additional, easy to measure, state variable. The presented approach makes use of this fact and uses the field winding as the receiver of a carrier signal, which is injected into the stator winding. Like this no magnetic saliency is mandatory in order to track the position. The new approach is compared to a well-known carrier signal based method, in which the stator works as the transmitter and as the receiver of the carrier signal. In this case a magnetic saliency is mandatory which differs from the saliency necessary for permanent magnet synchronous machines. The applicability of both approaches is proven in experiments.

Impact of iron losses on parameter identification of permanent magnet synchronous machines

Highly utilized permanent magnet synchronous machines (PMSMs) often show a significant inductance variation near and above the rated current values due to saturation. For precise modeling and control, the change in the parameters has to be considered by means of look-up tables. To parameterize these look-up tables, the inductances and the stator resistance have to be identified for various current set points. However, iron losses interfere with a precise identification if they are not considered. Therefore, an extended model is introduced to derive an identification method that considers and compensates for these effects. Measurement results prove the effectiveness of the proposed method.