WO2013063632A1

WO2013063632A1 - Method and device for detecting defects in magnetic slot closures of ac machines

Info

Publication number: WO2013063632A1
Application number: PCT/AT2012/000281
Authority: WO
Inventors: Thomas Wolbank
Original assignee: Technische Universität Wien
Priority date: 2011-11-03
Filing date: 2012-11-05
Publication date: 2013-05-10
Also published as: AT512058B1; AT512058A4

Abstract

In order to detect defects in magnetic slot closures of the stator of AC machines (2), preferably three-phase machines, in particular for detecting defects on slot closure wedges (34) or missing slot closure wedges (34'), the winding current (i) that is generated when a voltage (V) is applied to the machine is monitored, wherein the machine (2) is energised by means of a predetermined voltage signal (V) and the winding currents (i) generated thereby in the machine (2) are measured and evaluated in order to detect and localise stator-fixed asymmetries of the magnetic circuit arising from any defects based on the transient reactance of the machine (2).

Description

Method and device for detecting errors in magnetic groove closures of AC machines

The invention relates to a method and a device for detecting errors in magnetic Nutverschlüssen the stator of AC machines, preferably three-phase machines, in particular for the detection of errors in slot wedges or missing slot closure wedges, according to the preambles of the independent claims.

It is known to provide open stator slots in three-phase machines, in particular high-voltage three-phase machines, in order to facilitate the assembly of the stator windings. In order to prevent a decrease in the power factor due to the resulting air gaps, magnetic slot wedges are used, which are inserted after the mounting of the windings in the grooves and close them. Such slot wedges usually consist of a major proportion of iron powder and glass fibers and epoxy resin. The epoxy resin and the glass fibers prevent an electric line, whereby eddy currents can be prevented.

However, these keyway wedges may be damaged during operation of the machine and may also partially or completely disengage due to vibration, assembly failure and magnetic forces, thereby affecting the operation of the machine.

It is therefore important to detect such errors in magnetic Nutverschlüssen, as in particular the failure of Nutverschlusskeilen as early as possible in order to avoid any failure of the machine.

The problem with such magnetic slot wedges, as well as a suggestion for detecting errors in these magnetic slot wedges, is in the article Mike Davis, "Problems and Solutions with Magnetic Stator Wedges", Iris Rotating Machine Conference June 2007, San Antonio, TX (Mike DavisSmachinemonitor Specifically, it is proposed there for detecting errors in the magnetic slot closures, the operating currents of the three phases of the three-phase machine in Fre- to compare the spectrum of frequencies and to dissect the machine in case of noticeable deviations of the currents and to make a visual check. This optical inspection is presented as the most effective detection technique.

However, such an at least partial disassembly of the machine with the subsequent optical control is complicated and expensive, as can be seen immediately. Associated with such disassembly is also a relatively long failure of the machine, and it would be desirable to have a recognition and

Localization of defective magnetic slot closures

"Online", i.e. without disassembling the machine.

It is accordingly an object of the invention to detect errors in magnetic slot closures by measurements on closed machines and also to detect the position of these faults. It should also be possible to detect errors of magnetic Nutverschlüssen already in the initial state by such measurements on closed machines and thus to allow repair at an early, optimal time.

To solve this problem, the invention provides a method and a device as defined in the independent claims.

Advantageous embodiments and further developments are specified in the dependent claims.

In the present technique, predetermined voltage signals, in particular voltage pulses or high-frequency (sine) voltages, are applied to the machine, e.g. by means of a conventional converter, and the resulting winding current of the machine is measured in particular by means of current sensors usually present in the supply lines to the machine. The fault detection and localization is then based on the high-frequency or transient electrical properties of the machine. When high-frequency voltage signals or transient

Voltage signals (voltage pulses) are applied to the terminals of the machine, the resulting winding contains information about the magnetic state of the machine. In this current response, information about the properties of the magnetic material, but generally also about machine-inherent asymmetries, such as in the area of spatial saturation or of the grooves, are included as well as the desired error-induced asymmetries. Consequently, it is also a question of separating the possible inherent asymmetries from the desired error-induced asymmetries, and of evaluating only the latter with regard to the localization of the fault locations in the magnetic slot closures. Accordingly, currents obtained at different times and / or at different voltages are compared with each other or subtracted from each other to eliminate said machine-inherent asymmetries and to segregate and locate the stator-fixed asymmetries of the magnetic circuit originating from any errors.

Preferably, the voltage excitations and current measurements are performed in an unmagnetized state of the machine and / or in the unloaded state of the machine. This facilitates the measurement and localization of magnetic slot closure errors, if any. This advantage is enhanced when the detection is at a standstill machine

is carried out.

In order to efficiently perform the above-mentioned comparison of multiple current responses in the form of a subtraction, it is favorable if the voltage excitations and current measurements are simply performed at different successive rotor positions.

In the case that the machine is not inspected at standstill but with the rotor revolving, the influence of the resulting back EMF (EMF electromotive force) is eliminated by evaluating the current responses to various, in particular successive voltage excitations become.

Similarly, the influence of the stator resistance on the measurement can be eliminated by the current responses to different, in particular successive, voltage application. be evaluated.

For a rapid and effective evaluation of the measurement results, it is also advantageous if asymmetry pointers for a spectral analysis, e.g. by means of FFT.

In the case of three-phase machines, it is also advantageous if the same voltage excitations are carried out successively for all (three) phases.

As far as the present device is concerned, it is particularly advantageous for an efficient analysis of the transient processes when the converter is set up to deliver voltage pulses. Furthermore, it is favorable if the computing means for the derivation of asymmetry pointers are set up, wherein furthermore an advantageous embodiment is characterized by a spectral analysis unit for the spectral analysis of the asymmetry pointers.

The invention will be explained below with reference to particularly preferred embodiments, to which it should not be limited, and with reference to the accompanying drawings. In detail in the drawing:

1 shows schematically a representation of a groove with winding contained therein and a slot closure wedge.

2 shows schematically in the form of a block diagram an apparatus for detecting faults in magnetic slot closures of the stator of an electrical machine, wherein this machine, a three-phase machine, is assigned a converter for supplying power;

FIG. 3 shows a diagram of an example of a voltage pulse signal used to excite the machine and a corresponding current response, whereby sampling points are also illustrated; FIG.

4 shows in a diagram the direction-dependent inductance of a machine in the case of a missing slot closure wedge; 5 is a diagram of a pointer representation belonging to FIG. 4 for illustrating transient current changes for a given pulse sequence, for example in phase V;

Fig. 6A schematically shows a succession of grooves with windings and intact slot closure wedges and associated magnetic lines;

Fig. 6B shows a corresponding arrangement of these grooves with windings, but now missing in a groove of the slot closure wedge, resulting in irregularities in the magnetic line course;

Fig. 7 is a block schematic diagram illustrating the performance of the measurement and detection and associated signal processing, respectively;

8 shows a schematic diagram illustrating slot wedges of a machine, wherein tests have shown a slot closure wedge or a plurality of slot closure wedges removed; and

9 shows a diagram of the measurement results in a machine with an arrangement of slot closure wedges.

Electric machines whose stator windings are designed for high voltages are usually designed with open stator slots in which the windings can be easily accommodated; however, in order to avoid subsequently disturbances in the magnetic fields due to the open grooves, which could lead to power losses of the machine, the grooves are closed by magnetic sealing wedges. In Fig. 1, a part of a stator 30 is schematically illustrated with a groove 31 in perspective, wherein the groove is open to the inner periphery 32 of the stator 30 in itself, but after introduction of only schematically indicated in Fig. 1 stator windings 33 by means of an axially parallel inserted groove closure wedge 34 is closed.

This technique is well known, as well as it is known that the slot wedges 34 in the operation of the machine, if they have magnetic, but also mechanical forces, brennen etc., can loosen, can move in their guides and finally also break out of their guides and fall into the air gap. Such a loosening of the slot closure wedges may also have their cause in an insufficient assembly process. A loosened or, in particular, fallen-out slot lock wedge 34 can currently only be detected after at least partial disassembly of the machine. This is also confirmed by the cited reference by Mike Davis, where reference is made to the optical inspection, after suspicion has been ascertained, if applicable, about the current spectrum resulting in normal operation.

Otherwise, in the course of routine maintenance work, the determination of loosened or loosened slot wedges is made.

With the present technique, without dismantling the machine, errors in the area of the magnetic slot closures, in particular loosened, loosened or fallen slot wedges, can be detected, and, moreover, the location of the position of this magnetic fault is also possible. It is thereby possible to detect a reduced performance in time, and also in view of the possibility of recognizing the number of missing keyway wedges, it is facilitated the decision as to whether a machine must be disassembled or not.

In Fig. 2, a device 1 for detecting errors in magnetic Nutverschlüssen in an AC machine 2 is generally illustrated, with only a schematic example of a winding 3 of the machine 2 is indicated. By means of an inverter 4, the inputs of which a capacitor 5 is connected in parallel, voltage signals, namely concrete

Voltage pulses, as will be explained in more detail below with reference to FIG. 3, applied to the terminals of the machine 2. With the aid of the current sensors 6, 7 and 8, which are arranged in the supply lines (phases ü, V, W) from the inverter 4 to the machine 2, the resulting winding currents, hereinafter also called short current responses i, are detected and with the aid of a measuring unit 11 measured. This measuring unit 11 is closed by a cheneinheit 12 for evaluating the measured data, in particular for the calculation of an asymmetry pointer explained in more detail below, and moreover, a detection unit 13 for detecting the Rotornutung / position is provided. The units 11, 12 and 13 are combined to form a measuring and calculating system 14, and all three units are connected to a control unit 10, to which a start signal "Start" is fed via an input 19. This control unit 10 also controls one

Command unit 9 for the inverter 4 to corresponding

To issue switching commands for the machine excitation.

Furthermore, according to FIG. 2, the control unit 10 is connected to a data collection unit 15 and an analysis unit 16, preferably an analysis unit designed for an FFT (Fast Fourier Transformation) spectral filtering. These units 15 and 16 may be summarized as a final evaluation unit 18.

Finally, connected to the output of the analysis unit 16 is an error output unit, called error indicator 17 for short.

The present error detection and location technology is based, as mentioned, on a check of the transient reactance of the respective machine. For this purpose, for example, short voltage pulses are applied to the machine connections with the aid of the converter 4 shown in FIG. 2, and the resulting current is measured with the aid of the current sensors present in the individual supply lines. Instead of short voltage pulses and high-frequency voltage signals, in particular sinusoidal voltage signals, the machine 2 can be supplied. Incidentally, reference may also be made in this connection to the article by Fernando Briz et al., "Dynamic Operation of Carrier-Signal-In- formation-Based Sensorless Direct Field-Oriented AC Drives", IEEE Transactions on Industry Applications, 2000, p. 9994 (00) 07622-2, where harmonics, which are caused by transient phenomena of the fundamental wave currents, are examined in order to be able to estimate, inter alia, a rotor position with high accuracy.

As mentioned, in the present technology, one item is the Identification of the transient reactance or impedance of the respective machine. In the following, the theoretical principles will be explained in more detail in this connection by way of example for the embodiment with the excitation of the machine with the aid of short voltage pulses. In Fig. 3 for this purpose, a pulse-shaped voltage signal V is schematically shown in conjunction with the then resulting in the machine current i. The pulse sequence has a positive voltage pulse I and a negative voltage pulse II. The current i has as shown a triangular course. The detection of two time derivatives of the current response i is accomplished with, for example, four sampling points S ₁ -S 4, the current values at the sampling points S _x , S _{2 being} the slope of the current increase and the sampling values S ₃ , S _{4 being} the angle of the current waveform in FIG let calculate the negative pulse phase II. Accordingly, the actual current values are measured a short time after the start and a short time after the end of each pulse period. From this, the time derivatives of the current can be approximated. The entire measurement can be done within a few 10 ps.

The time profile of the current signal i is essentially determined by the transient leakage inductance of the machine. An essential part of the stray flux is the stator slot leakage flux and the zigzag stray flux. In a symmetrical configuration of the stator leakage flux passing through the slot closure wedges ^¬ independent of the spatial angle of the high frequency or transient excitation. However, if a slot wedge is missing or multiple slot wedges are missing, there will be a significant change in the transient leakage paths as the excitation angle changes. These changes lead to an asymmetry that can be detected by the measuring process. In the present technology, the measurement is preferably used with the machine at rest, at zero load and at zero flow, ie preferably when the machine is not being operated or is currently being serviced.

By way of example, an electric machine 2 with three phases U, V, W is assumed as well (ie a three-phase machine), which is supplied with inverter, wherein the various Voltage states can be used at the inverter outputs. When the output of the inverter 4 changes from the inactive state to an active state, a voltage jump corresponding to the amplitude of the intermediate voltage of the converter 4 is generated. The same applies to the change from the active initial state to the inactive one.

As mentioned, the current measurement is carried out with the aid of the built-in, usually present current sensors 6, 7, 8. When these measurements are taken at a zero flux and a zero load, the power of the inverter 4 may be a fraction relative to that of the machine 2.

In an assumed symmetrical machine 2, a voltage space vector, which describes the behavior of the machine, is known to result according to the following relationship (1):

Here the individual designations have the following meaning: Voltage-space vector r _s Stator resistance is stator current

// stray inductance back EMF.

The voltage pointer caused by the inverter switching action leads to the aforementioned current change di _s l dr. This deflection of the current with time is determined by the switching state of the inverter 4 from the DC intermediate voltage, but also by the stray inductance //, from the stator resistance r _s and influenced by the back EMF d \ _R ldr. After realizing the voltage level (some ΙΟμβ long), the transient response of the engine 2 by the voltage drop of the transient leakage inductance / _/ ,, is dominated. An additional influence exists due to back EMF if the machine 2 is not tested at standstill. In the latter case, identification of the transient leakage inductance is possible only after elimination of the disturbance due to the back EMF.

The influence of the back EMF as well as the stator resistance voltage drop can be easily eliminated by evaluating the current responses to two voltage jumps with different inverter output states. When these two output states follow each other, the fundamental operating point and in particular the direction and magnitude of the back EMF do not change significantly. These two successive inverter output states are the phases (intervals) I and II in the diagram of Fig. 3. In the following, the quantities are also given in the further theoretical considerations in the relationships with the additional index I and II depending on the phase. Accordingly, to eliminate the spurious components, a simple subtraction of the two stator equations (see equation (1) above) can be performed. Furthermore, the transient inductance of the machine differs from the fundamental leakage flux and is designated t. This results in the following equation for the resulting voltage vector:

The fundamental operating point of the stator current is as well as the time derivative thereof can be measured directly. Even if, as mentioned, the machine 2 is not operated at standstill, the back EMF does not change significantly between the two consecutive pulse pickups. Accordingly, in subtracting the two stator voltage equations, the back EMF precipitates. The same applies to the influence of the stator resistance. If the voltage phasors of the successive pulses point in opposite spatial directions, then the fundamental operating points in the two cases are practically the same due to the symmetrical pulse pattern as shown in FIG.

If an ideal symmetrical machine 2, with operation at one Zero main flux and a zero speed, the transient leakage inductance In may be considered as a scalar quantity, and accordingly the direction of the slope of the resulting current is parallel to the excitation pulse voltage. In current machines, however, even if they have no errors, there are always inherent spatial asymmetries, which also lead to an angular dependence of the transient leakage flux.

Therefore, the directions of the voltage pulses and the resulting current slopes are not equal. For this purpose a complex transient leakage inductance { _{it can be} introduced, cf. the following equation:

(3)

In this way, the asymmetries of the three phases are combined in one parameter. This complex transient leakage inductance is composed of a scalar offset value 1 _{0 // *} * and a complex value I "od, as seen in equation (3). The offset value represents the symmetric machine, whereas the complex value indicates the error-induced asymmetry with magnitude and spatial direction. It is assumed here that there is no further asymmetry. y is the angle of the asymmetry part, ie it indicates the spatial position of the maximum inductance within a pole pair. The asymmetry therefore has a double period with respect to a single electrical revolution and a fixed position relative to the stator.

In this way, in the case of excitation of the machine 2 by means of a voltage pulse sequence (see Fig. 3) in the evaluation of the current response, the time derivative of the machine current into a symmetric part and an asymmetric part, which for the present problem of Interest is to be separated.

The size and angle of the complex inductance part contain the information concerning the error-caused asymmetry. In the case a missing slot wedge corresponds to its spatial position of the direction of the complex inductance relative to the stator windings.

For the present evaluation, it suffices to monitor the resulting current bias of the two switching states after eliminating the stator resistance and, if appropriate, the back EMF. For the calculation of the angle y of the asymmetry, an exact knowledge of the scalar value and the complex value of the inductance is required. Accordingly, the above equation (3) is substituted into the equation (2), and the resulting equation is inverted, resulting in the following equation (4), in which the complexed-complex quantities are indicated with a star:

^ λ - ^ ir = yoffse, ^■ s. - -s, II) + _mo A, iy.s, n) ()

This equation can be simplified as follows:

Δτ S, I - (5)

It is thus sufficient that the measuring system simply calculates the simplified equation (5), wherein the time derivatives are numerically calculated by difference quotients, as is known per se. A corresponding index change takes place at the voltage pointers.

The measured current slope of the excitation sequence is thus directly influenced by the size and position of the complex inductance part according to equation (4); the values of y ₀ / fsei and _ $ can be obtained after conversion as indicated in equations (6): y offset -p.

(6]

The main direction of the resulting current change is determined by the two converter switching states (I, II in Fig. 3), as can be seen from the above equations. In symmetric machines, the scalar value y ₀ ff _Se i, while the size of the complex value, usually dominates

even at one

error-prone machine is only about 10% of this scalar offset value. FIG. 4 illustrates the relationships described above in the case of an asymmetry caused by a missing slot closure wedge in the stator in an angular position γ = 50 °. The missing wedge at the position γ = 50 ° is designated 34 'and crossed out to illustrate its absence. The resulting asymmetry leads, for example, to an ideal sinusoidal modulation of the stray inductance. The corresponding directional dependence of this inductance is indicated by an ellipse 35 with the minimum inductance in the direction of the magnetic axis corresponding to a magnetomotive force in the groove where the wedge is absent. The directional inductance is illustrated in FIG. 4 with the pointer 36. A similar ellipse is evidently obtained if a slot lock key is missing in the position y = 230 °, as also results from the equations (3) and (6).

The corresponding signals are schematically illustrated in the phasor diagram of FIG. 5, the differential voltage phasor Vj -ii being in the phase V direction

shows. The resulting current change pointer A i_ _{s I} _ "/ Δ τ is composed as described of two parts, namely the scalar part s, in.yoffset and the complex part j -R _xm0 d, see. also the above equations (4) and (5). The position of the asymmetry is selected similar to that in FIG. 4 in FIG. 5, the main axis of the ellipse being illustrated as a pointer 37 for the maximum inductance in FIG. 5. The angle δ defines the difference of the direction of maximum transient inductance (slot wedge) relative to the direction of the differential voltage vector Vs, / - // of the pulse excitation.

In simple terms, the error-induced asymmetry leads to a current slope, arrow 38 in FIG. 5, which is composed of the scalar value parallel to the direction of the excitation (phase V) in the example shown, arrow 39, and the complex component, pointer 40 in FIG. 5, with a direction corresponding to twice the angle δ. If the direction of asymmetry is equal to the direction of the resulting excitation voltage (δ = 0 °), then the resulting current slope will have the same direction as the applied resultant voltage if the machine were symmetric. However, compared to a symmetrical machine, the size is smaller. Again, with an angular difference δ = 90 °, a "symmetrical" direction is obtained, but now with a larger size As can be seen, the tip of the resulting pointer 38 moves twice to the left of the circle 38 'as the position of the missing slot lock key 34' moves an electrical period changes.

In order to analyze the described influence of a missing slot closure wedge 3 'on the current change and to find confirmation for the above explanations, a simulation based on finite elements (FE) was carried out, the result being illustrated schematically in FIGS. 6A and 6B. In particular, FIG. 6A shows part of a stator with grooves, all of which are properly closed by slot wedges. This results in an image of the magnetic lines as shown.

In Fig. 6B then a missing slot closure key 34 'is given. This results in disturbances or changes in the course of the magnetic lines, as shown in particular with the three located on the right side in Fig. 6B arrows 41.

The machine 2 was operated without fundamental wave excitation. Only voltage pulses were applied in the direction of the phase axis U (see also FIG. 8, which will be described in greater detail below). The remote slot wedge 34 'was in a groove of the phase winding U (electrically at right angles to the phase axis). The changes 41 in the course of the magnetic lines are directly linked to the change in the transient current slope, as determined by the measurements. This eventually leads to a fixed stator asymmetry, which is detectable in the asymmetry pointer.

In practice, the controls and measurements can be done with the aid of a processor, the voltage signals for the excitation and causes trigger signals for the measurement. The voltage-pulse sequence can be generated from stored values. Accordingly, the voltage pointer 39 (Figure 5) is predictable in advance. The current difference pointer 38 results from the current samples S 1 to S ₄ taken in the respective case (compare FIG. 3). Since this current difference pointer is composed of a symmetrical part 39 and an asymmetrical part 40, these two parts are separated from each other in the measuring signal. This can be accomplished, for example, as follows: As can be seen from FIG. 5, the part influenced by the scalar value y ₀ ffsa (pointer 39 in FIG. 5) points in the direction of the excitation pulses. The voltage pulse sequences are applied successively in the three main phase directions. If the three resulting current difference hands are combined by addition, this results in only one current difference pointer. In this pointer calculation, the symmetric components are eliminated. The resulting pointer now contains more information about the machine 2 asymmetries. This pointer is referred to simply as an asymmetry pointer.

For the final detection of missing (or defective) slot wedges, a few more signal processing steps are performed. The obtained asymmetry pointer contains not only the error-induced asymmetry but also some machine-inherent asymmetries. The essential machine-inherent asymmetry is caused by spatial saturation, by groove, and by rotor anisotropy. These asymmetries exist in virtually all machines, even if they are otherwise flawless. These asymmetries are all superimposed in the asymmetry pointer, but identification and separation can be performed according to their deterministic behavior. The detected asymmetry pointer is thus split into these components.

An essential machine-inherent asymmetry in induction machines is the saturation asymmetry. This is due to the different levels of saturation caused by the

Basic wave along the circumference of the machine is effected. The modulation period is twice the period of the fundamental electric wave corresponding to the number of poles. The Identification and separation of this asymmetry can be carried out in advance.

Another asymmetry is caused by the openings of the grooves in the lamination. The period of this asymmetry is related to the rotor angle. In machines with magnetic slot closures, however, their modulation is very small, and this asymmetry can usually be neglected.

FIG. 7 illustrates, in a type of block diagram, the signal processing in the present device 1, wherein in particular the elimination of the machine-inherent asymmetries is also included.

According to FIG. 7, for the measurement and evaluation of the measurements, a preferably uniform processor system 50 with a

Control unit 10 'for the individual processes and with a measurement and calculation system 14' is provided. In the course of the measurement, in the present example voltage pulses are emitted by a unit 4 '(for example the converter 4 according to FIG. 2), and in the actual measuring unit 11' (see the current sensors 6, 7 and 8 as well as the unit 11 in FIG. 1), the measurement of current responses is as described above. From these measurement signals, the asymmetry pointer is calculated as explained above with the aid of a computing unit 12 '(see also the block 12 concerning the asymmetry-pointer calculation in FIG. Specifically, the voltage pulses are output from the inverter 4, the currents in the individual phases U, V, W are measured by means of the sensors 6, 7 and 8, in accordance with FIG. 3 to the respective sampling points Sx to S. In the arithmetic unit 12 ', the current differences are calculated and stored in accordance with the above equations (4) and (5).

The generation of the trigger signals for the switching of the converter 4 and for the taking of the measurements is accomplished by the control unit 10 or 10 ', which may be realized on the basis of a real-time processor, preferably together with the computing components of the system 50.

The same kind of pulse sequences are successively applied to all three phases. sen U, V, W, and as mentioned, by combining the computed pointers for the three directions (phases), the symmetric part is eliminated and the desired asymmetry pointer is obtained.

Next, with the aid of a rotor drive unit 51, the position of the rotor of the machine 2 is changed and the voltage signal excitation, measurements and calculations are repeated.

This process is repeated for at least one rotor grooving period or until the rotor has performed a mechanical rotation.

Data storage with respect to the calculated asymmetry pointers takes place in a memory unit 15 '.

In a spectral analysis unit 16 ', for example for an FFT analysis, the asymmetry pointers are then analyzed, and the notching modulation is also identified and eliminated. If there is a saturation asymmetry, e.g. According to the magnets in a permanent magnet machine, these can also be eliminated, for example, by spectral filtering.

The result regarding the error display is finally displayed or output via a unit 17 '.

Since the asymmetry pointer is a complex size as mentioned above, the input to the analysis unit 16 '(or 16 in Fig. 2) contains size and direction information. As a result, it is also possible to detect both single and several missing slot closure wedges 34 'together with their positions.

Finally, we will discuss practical tests carried out with a four-pole 11 kW induction machine, in which first a slot closure wedge and then further slot closure wedges were removed. The measurements were carried out in the state without load and without flux and without magnetization of the machine. Preliminary measurements were made on the symmetrical machine, where all slot wedges were still present, to obtain reference values for the fault indicator system.

Thereafter, various measurements were made after removal of individual slot lock wedges, first removing a slot lock wedge in position A (also designated 34 'in Fig. 8, corresponding to Fig. 4). With dashed borders, the wedges for the phase U are indicated in FIG. 8, the axis of the U phase assuming the

horizontal axis in the diagram of FIG. 8 correspond. In other words, the wedges U according to the borders 60 correspond to the windings active in the phase U.

As mentioned above, removal of a slot closure key in the + 90 ° position, as shown in Figure 8 (but also the wedge in the -90 ° position), results in a change in the error indicator in the direction of the phase axis U. Even if only the absence of the slot closure wedge 34 'in the position + 90 ° (or -90 °) leads exactly to the phase direction U, the absence of slot wedges in positions +/- 70 ⁰ and +/- 110 ⁰ (see

Positions B and C in Fig. 8) are nonetheless referred to as "missing slot wedge in phase U".

In the diagram of Fig. 9, imaginary part and real part (each in arbitrary units (abitrary units) are illustrated in a phasor representation, also showing the phases U, V, W of the engine 2. The obtained measurement results are shown in Fig. 9 follows registered:

(1) Item 61 - flawless machine, i. all slot closure wedges 34 are present;

(2) Item 62 - the slot lock key 34 'in the position A (see Fig. 8) has been removed;

(3) Item 63 - here the slot wedges in positions A and B have been removed; As can be seen, this increases Asymmetry; it is of course similar, when the wedges in the positions A and C are removed as shown in FIG 8;

(4) Item 64 - the slot wedges in positions A, B and C have been removed; an even greater asymmetry results, as can be seen directly from FIG. 9;

(5) Item 65 - here, in addition to the slot wedges in the positions A, B and C, those in the position D (see Fig. 8) were also removed, that is, a wedge in a position outside the phase U, but rather the phase W associated;

Comparing the various measurement results in FIG. 9, it can be seen that each missing keyway key is correlated to a pointer of the error indicator. The combination with several missing slot wedges results from summation of the individual pointers. It is not only possible to recognize missing slot closure wedges 34 'as such, but also to distinguish between individual or multiple missing slot wedges and also determine their spatial positions.

Claims

Claims:

1. A method for detecting errors in magnetic Nutverschlüssen of the stator of AC machines (2), preferably three-phase machines, in particular for detecting errors in Nutverschlusskeilen (34) or missing Nutverschlusskeilen (34 ¹ ), in which method when applying a voltage (V) to the machine generated (2) winding current () is checked, characterized in that the machine (2) with a predetermined voltage signal (V) is excited and thereby in the machine (2) generated winding currents () measured and evaluated to detect and locate stator-fixed asymmetries of the magnetic circuit based on the transient reactance of the machine (2) that originate from any errors.

2. The method according to claim 1, characterized in that the machine (2) with voltage pulses (V) is excited.

3. The method according to claim 1, characterized in that the machine (2) is excited by a high-frequency voltage.

4. The method according to any one of claims 1 to 3, characterized in that the detection is performed when the machine is stationary (2).

5. The method according to any one of claims 1 to 4, characterized in that the voltage excitations and current measurements are carried out at different rotor positions.

6. The method according to any one of claims 1 to 3, characterized in that when the machine (2) is examined with rotating rotor, the influence of the thereby caused back EMF is eliminated in that the current responses () to different , in particular successive voltage excitations are evaluated.

7. The method according to any one of claims 1 to 6, characterized in that a possible influence of the stator resistance on the measurement is eliminated in that the current responses () to different, in particular successive voltage application be evaluated

8. The method according to any one of claims 1 to 7, characterized in that the voltage excitations and measurements in a non-magnetized and / or unloaded state of the machine (2) are performed.

9. Method according to one of claims 1 to 8, characterized in that from the measurements asymmetry pointer (40) for a spectral analysis, e.g. by means of FFT.

10. The method according to any one of claims 1 to 9, characterized in that in the case of three-phase machines, the same voltage excitations are carried out successively for all phases (U, V, W).

11. Apparatus (1) for detecting errors in magnetic slot closures of the stator of AC machines (2), preferably three-phase machines, in particular for detecting defects in slot wedges (34) or missing slot closure wedges (34 '), with means, e.g. a converter

(4), for generating voltages (V) for application to the machine (2), characterized in that in the supply lines to the voltage terminals of the machine (2) current sensors (6, 7, 8) for measuring the current Reactions () of the machine (2) are arranged on the voltage excitations (V), using computational means

(14, 18) arranged to evaluate the resulting currents to detect and locate stator-fixed asymmetries of the magnetic circuit derived from any errors based on the transient reactance of the machine.

12. The device according to claim 11, characterized in that the converter (4) is arranged for the delivery of voltage pulses (V).

13. Device according to one of claims 11 or 12, characterized in that the computing means (14, 18, 14 ', 15') are arranged for the derivation of asymmetry pointers (40).

14. The apparatus according to claim 13, characterized by a Spectral analysis unit (16; 16 ') for the spectral analysis of the asymmetry pointers (40).