Juergen L. Machl

Combination of signal injection and neural networks for sensorless control of inverter fed induction machines

For mechanical sensorless control of inverter-fed induction machines, a satisfactory performance at low speed down to zero fundamental frequency can so far only be achieved by evaluating inherent saliencies of the induction machine. Similar to other sensorless methods based on signal injection, the resulting control signals of the indirect flux detection method by on-line reactance measurement is influenced by every saliency, for example, the saturation based, the slotting, and the anisotropy saliency as well as by load and flux level. Since these influences are extremely dependent on the machine design, they can hardly be calculated in advance and removed by filtering or digital signal processing. However, the possibility of utilizing a neural network for learning the individual dependencies and removing the unwanted influences can provide a very satisfactory result. Since the easy implementation of a neural network does only use a small amount of calculation power, the algorithms can be implemented even in low-cost signal processors. Measurements on mechanical sensorless controlled induction machines present adequate results up to about rated load, depending on the transient electrical behaviour, and with this on the design parameters of the induction machine.

Closed-loop compensating sensors versus new current derivative sensors for shaft-sensorless control of inverter fed induction machines

The technology of shaft-sensorless control of ac machine has been developed through the past decade. Some improvements have been made on the evaluation of position information out of so-called nonmodel based methods, which do not have the drawback of the mathematical models. However, the improvement in the field of the current sensors was not focused on mechanical sensorless control and, thus, has not brought that much enhancement for this particular topic. With the introduction of the current derivative sensor (CDI) a new concept specialized for sensorless control has been proposed. The quality of the control signals can be increased, and the drawbacks with the measurement sequences needed for the detection cycles are reduced with the help of inexpensive and easy-to-use derivative current sensors.

Influence of inverter-nonlinearity and measurement setup on zero speed sensorless control of AC machines based on voltage pulse injection

In this paper the influence of the non-ideal behavior of power electronics, sensors, and signal processing on speed sensorless control is addressed and investigated. The speed sensorless control signals are obtained from the transient current response resulting from an excitation of the machine with voltage pulses. The investigations are focused on sensorless control of induction machines, however the results are applicable also for other types of AC machines. Reference to the inverter the influence of interlock dead-time as well as the voltage drop of the power electronics is assessed. Simulation results are given to support the separation of the different effects. Additionally measurement results are presented to show the influence on the resulting control signals. Different ways to consider and compensate the different deteriorating effects with respect to sensorless control are discussed and compared.

Lamination material anisotropy and its influence on the operation of inverter-fed induction machines

In modern inverter-fed machines the exploitation of the transient electrical behavior becomes of increasing interest to obtain additional information on the machine state with speed sensorless flux/rotor position detection schemes or fault condition methods. In the application of these methods it turned out that the anisotropy of the lamination material, which is usually negligible in the fundamental wave behavior of the machines influences the transient control signals and may act as a disturbance during operation. In this paper the anisotropy is determined by hysteresis and torque measurements of lamination sheet as well as by the evaluation of the transient electrical behavior of industrially manufactured inverter fed induction machines.

Prediction and measurements of a current derivative sensor response for voltage pulses applied to induction machines

Current derivative sensors (CDI’s) can advantageously be used in modern drives to determine the position of the rotor or flux in induction machines if no sensor at the shaft is available or to realize an effective condition monitoring of the drive. This is achieved by exploiting the transient current slope of the machine due to voltage pulses impressed by the switching of the inverter. To enable a CDI optimization with reference to the demands given by the drive application, a simulation model is presented which describes the nonlinear dependence on the history of the magnetic field and magnetization in ferromagnetic materials by statistical domain behavior using phenomenological parameters. In addition, measurements and simulations of the developed sensor output signals are presented which are used in a 15 kW induction machine drive. The optimized CDI sensitivity is S=0.35 μΩs and the bandwidth exceeds the characteristic of standard current sensors by five times.

Zero speed sensorless control signals of induction motors with closed rotor slots

Realising control of standard induction machines at low and zero speed without mechanical sensors implies the use of parasitic nonfundamental wave effects. Different schemes have been proposed in the past which exploit the inherent saliencies of saturation and/or rotor slotting to detect the main flux and the rotor position. They all use a high frequency or transient signal in addition to the fundamental wave excitation which is impressed by the inverter. The machines reaction on this high frequency excitation is measured and the flux and/or rotor position signal is extracted by special algorithms. However, this transient electrical behaviour and the resulting control signals of an induction machine are strongly influenced by the design of the machine and especially the shape of the lamination. The two most prominent saliencies present in standard induction machines are caused by saturation and slotting, enabling thus either sensorless flux or rotor position detection. Especially the rotor position detection suffers from a strong dependence on the slot geometry of the lamination. Yet the details for this influence are not quite clear and have not been addressed in literature. In this paper measurements were performed on machines giving a comparison of closed and opened rotor slots. With the help of measurement windings placed along the air gap the spatial distribution of the transient leakage flux caused by the high frequency excitation is determined to make an interpretation of the transient behaviour more accurate. The measurements made in this paper are based on a transient excitation of the machine with voltage pulses. The results are however also applicable to high frequency sinusoidal excitations.

Simulation and observer based detection of airgap asymmetries caused by rotor eccentricity in inverter fed AC machines

A new method to detect fault conditions in inverter fed AC machines caused by asymmetries in the airgap is presented in this contribution. In the first part the influence of these asymmetries on the transient electrical behavior of the machine is investigated by simulation results obtained from a magnetic equivalent circuit model. Starting from these results the proposed method is derived. It is based on the reaction of the machine current to voltage pulses caused by the switching of the inverter. The continuous switching of the inverter is necessary for the current control loop during normal operation of the drive. Using a special observer structure any changes in the reaction of the machine caused by airgap asymmetries can be detected and isolated. The only inputs necessary for the realization of the method are the DC link values of the voltage and the current. The observer estimates the transient parameters of the three phases and by comparing the three phase values an error signal can be calculated. The method is insensitive to changes of the machine parameters due to temperature or saturation as long as these changes occur symmetrically in all three phases. The method is applicable in both steady state and transient operation. The practical operation of the method is verified by measurement results at different points of operation and with different fault positions/magnitudes.

Lamination Design Variations for Improved Performance of Zero Speed Sensorless Controlled Induction Machines

Abstract To realise a high-dynamic controlled operation of induction machines the flux position has to be estimated during operation of the drive. Omitting the shaft sensor leads to a deterioration of the performance at low fundamental frequencies if fundamental wave models of the machine are applied. To determine the flux position at zero speed without shaft sensor it is thus necessary to use parasitic, non-fundamental wave effects of standard induction machines, such as spatial saturation, slotting, or other anisotropy. These effects are not evident in normal operation but can be exploited using the high frequency or transient electrical behaviour of the machine. All sensorless zero speed schemes currently published thus make use of a high frequency or transient excitation of the machine in addition to the fundamental wave, which are both impressed by the inverter. The machine reaction on this high frequency excitation is measured and the flux and/or rotor position signal estimated using special algorithms and signal processing techniques. However, it turned out in the past that the shape of the lamination and especially the slot geometry have strong influence on the high frequency/transient electrical response of the machine. Before realising a sensorless controlled drive it is thus advantageous to have a look at the design of the machine as not any design is suitable for a specific sensorless control algorithm. The presented investigation is thus focused on the influence of the lamination geometry on the resulting signals of a sensorless control scheme. The sensorless control method applied is usually denoted as INFORM-method in literature and is based on a transient excitation with voltage pulses. In the transient electrical current response of the machine there are always all saliencies present. The two most prominent saliencies are caused by saturation and slotting. Their modulations in the control signals have to be extracted and separated in order to obtain the flux or the rotor position of the machine. Currently the interaction of the saliencies limits the practical application of these methods. To investigate the mentioned influence, measurements have been performed on five machines with different lamination geometry By comparing the extracted saliencies of these machines, the influence of slot design on the sensorless control signals is depicted. Thus the magnitudes of the two saliencies can be influenced by a proper adjustment of the geometry This association is shown for the saturation saliency in Fig. 1. Depending if the flux or rotor position is to be used for the machine control the corresponding saliency can be boosted and the others attenuated. This results in a more reliable extraction and separation ofthe saliencies and a more robust control.