Method for determining frequency-dependent winding impedance and drive for carrying out a method according to any of the preceding claims

Abstract

The invention relates to a method for determining frequency-dependent winding impedance, wherein, in succession for different first frequencies fn, a sinusoidal voltage is applied to at least one of the motor phases, the frequency of the sinusoidal voltage being the particular first frequency fn, wherein the values ​​of the motor current detected by a current sensor are converted by a sigma-delta modulator into a digital data stream which: - is supplied to a first decimation filter; - for a plurality of natural numbers k, is multiplied by a time-dependent sinusoidal signal in each case, the frequency of the sinusoidal signal being the particular first frequency fn multiplied by k in each case, and a moving average of the product thus created is formed to determine each first spectral component; and - for the natural numbers k, is multiplied by a time-dependent cosinusoidal signal in each case, the frequency of the cosinusoidal signal being the particular first frequency fn multiplied by k in each case, and a moving average of the product thus created is formed to determine each second spectral component, wherein the winding impedance Z is determined from the spectral components of the current and the voltage for each of the first frequencies and their k-multiplied multiples.

Description

[0001] Method for determining the frequency-dependent winding impedance and drive for carrying out a method according to one of the preceding claims

[0002] Description:

[0003] The invention relates to a method for determining the frequency-dependent winding impedance and a drive for carrying out a method according to one of the preceding claims.

[0004] It is well known that the stator winding of an electric motor has a winding impedance.

[0005] From CN 1 15 166 515 A, the closest prior art discloses a method for determining the insulation state of a stator.

[0006] From US 6 483 319 B1 a method for performing a broadband impedance response to predict a failure of a stator winding is known.

[0007] From WO 2016 / 029 234 A1 a method for detecting the condition of an insulation of an alternating current machine is known.

[0008] From WO 2022 / 094 726 A1 a determination and classification of a deterioration of the insulation of a stator winding of an electric motor is known.

[0009] The invention is therefore based on the object of further developing the operational reliability of an electric motor, in particular whereby condition monitoring should be applicable.

[0010] According to the invention, the object is achieved by the method according to the features specified in claim 1 and by the drive according to claim 11.

[0011] Important features of the invention in the method for determining the frequency-dependent winding impedance of a three-phase motor fed by an inverter or converter having a current sensor are that, in particular during commissioning of the three-phase motor, successively in time intervals for different first frequencies f n in each case a time-dependent, in particular rectangular, voltage waveform, the frequency of which is the respective first frequency f nis applied to at least one of the motor phases, wherein the values ​​of the motor current detected by a current sensor, in particular analog, are converted by a sigma-delta modulator into a digital data stream, which is fed to a first decimation filter to form a moving average, in particular which is used as a spectral component at vanishing frequency, is multiplied for several natural numbers k in each case by a time-dependent sinusoidal signal, the frequency of which is the respective k-fold of the respective first frequency f n and a moving average of the resulting product is formed to determine a respective first spectral component and for the natural numbers k is multiplied by a time-dependent cosinusoidal signal whose frequency is the respective k-fold of the respective first frequency f nand a moving average of the resulting product is formed to determine a respective second spectral component, wherein the winding impedance Z is determined from the spectral components of the current and the voltage for each of the first frequencies and their k-fold multiples, in particular by quotient formation, in particular according to jF J

[0012] “ / n ), in particular where U(f n ) the spectral component of the voltage at the first

[0013] Frequency f n is and I (f n ) the Fourier component of the current at the first frequency f n is.

[0014] The advantage here is that a rectangular voltage waveform with a first frequency is first applied, and the spectral components of the current are determined for the resulting current waveform. Thus, the real and imaginary components of the Fourier transform of the current at this first frequency are known. However, since the spectral component of the applied rectangular voltage waveform is also known or can be easily determined for this first frequency (the voltage waveform does not need to be recorded using sensors, but the amplitude of the voltage waveform is already known), the impedance can be easily determined as the quotient of this spectral component of the voltage at the first frequency and the Fourier transform of the current at the first frequency.It is important that the recorded current value is first converted into a 1-bit data stream by a sigma-delta modulator. The spectral components can then be determined using a simple decimal filter, provided the product of the 1-bit data stream and a sinusoidal or cosinusoidal signal with the first frequency is applied to its input. In subsequent steps, the frequency of the first sinusoidal or cosinusoidal signal is increased, namely by a factor of k, and the impedance for this respective frequency is again determined. This is performed for several integer values ​​of k, for example, for k = 1, then for k = 2, then for k = 3, etc.

[0015] In this way, impedance values ​​can be determined for the given values ​​of the first frequency and their integer multiples. The set of first frequencies used is f nand natural numbers k is chosen so that the impedance in the frequency domain can be represented with at least 30 points.

[0016] This allows the impedance to be monitored for the occurrence of antiresonances, which can arise due to voltage stress in the stator winding. For this purpose, the impedance, particularly the magnitude of the impedance, can be monitored either for excessive deviations from a specified value or frequency-dependent value profile, or a local minimum of the impedance in the frequency domain can be determined to monitor for emerging antiresonances. This enables condition monitoring of the electric motor, and the reliability of its operation can be improved.

[0017] Another advantage of the invention is that the semiconductor switches present in the converter or inverter, intended for controlled operation, provide the square-wave voltage of the respective phase winding. Particularly during commissioning, there is sufficient time to measure each motor phase sequentially in the manner described and thus to determine the winding impedance of each of the three motor phases, in particular phase windings, separately. Furthermore, the current measurement of the converter or inverter, intended for controlled operation, can be used according to the invention. Thus, no additional hardware expenditure is necessary to implement the invention.

[0018] In a further development, only a single decimation filter can be used, to which the current waveform multiplied by the sinusoidal or cosinusoidal signal is fed several times in succession to determine the respective spectral component at the respective first frequency or k times the respective first frequency. For k = 0, even simple averaging is obtained, so that only a single decimation filter is sufficient overall. The disadvantage of this is that the discrete data stream of the moving average on the output side is only available at long time intervals.

[0019] Important features of the method for determining the frequency-dependent winding impedance of a three-phase motor fed by an inverter or converter having a current sensor according to claim 2 are that, in particular during commissioning of the three-phase motor, successively in time intervals for different first frequencies f n in each case a time-dependent, in particular rectangular, voltage waveform, the frequency of which is the respective first frequency f n is applied to at least one of the motor phases, wherein the values ​​of the motor current detected by a current sensor, in particular analog, are converted by a sigma-delta modulator into a digital data stream, which is fed to a first decimation filter, a first multiplier, which for a plurality of natural numbers k each carries out a multiplication with a time-dependent sinusoidal signal, the frequency of which is the respective k-fold of the respective first frequency f nand a second multiplier which carries out a multiplication for each of the plurality of natural numbers k with a time-dependent cosinusoidal signal whose frequency is the respective k-fold of the respective first frequency f n is supplied, wherein the output signal of the first multiplier is supplied to a second decimation filter for determining a respective first spectral component, wherein the output signal of the second multiplier is supplied to a third decimation filter for determining a respective second spectral component, wherein the winding impedance Z is determined from the spectral components of the current and the voltage for each of the first frequencies and their k-fold frequencies, in particular by quotient formation, in particular according to in particular where U(f n ) the spectral component of the voltage at the first frequency f n is and l(f n) the Fourier component of the current at the first frequency f n is, in particular wherein the respective spectral component of the voltage is determined from the voltage curve applied by the inverter or converter, in particular mathematically.

[0020] The advantage here is that a rectangular voltage waveform with a first frequency is first applied, and the spectral components of the current are determined for the resulting current waveform. Thus, the real and imaginary components of the Fourier transform of the current at this first frequency are known. However, since the spectral component of the rectangular voltage waveform is also known or can be easily determined for this first frequency (the voltage waveform does not need to be recorded using sensors, but the amplitude of the voltage waveform is already known), the impedance can be easily determined as the quotient of this spectral component of the voltage at the first frequency and the Fourier transform of the current at the first frequency.It is important that the recorded current value is first converted into a 1-bit data stream by a sigma-delta modulator. The spectral components can then be determined using a simple decimal filter, provided the product of the 1-bit data stream and a sinusoidal or cosinusoidal signal with the first frequency is applied to its input. In subsequent steps, the frequency of the first sinusoidal or cosinusoidal signal is increased, namely by a factor of k, and the impedance for this respective frequency is determined. This is performed for several integer values ​​of k, for example, for k = 1, then for k = 2, then for k = 3, etc.

[0021] In this way, impedance values ​​can be determined for the given values ​​of the first frequency and their integer multiples. The set of first frequencies used is f nand integers k is selected such that the impedance can be represented in the frequency domain with at least 30 points. This allows the impedance to be monitored for the occurrence of antiresonances, which can arise from voltage stress in the stator winding. For this purpose, the impedance, in particular the magnitude of the impedance, can either be monitored for excessive deviations from a specified value or frequency-dependent value curves, or a local minimum of the impedance in the frequency domain is determined to monitor for emerging antiresonances. This enables condition monitoring for the electric motor, and the reliability of its operation can be improved.

[0022] Another advantage of the invention is that the semiconductor switches present in the converter or inverter, intended for controlled operation, provide the square-wave voltage of the respective phase winding. Particularly during commissioning, there is sufficient time to measure each motor phase sequentially in the manner described and thus to determine the winding impedance of each of the three motor phases, in particular phase windings, separately. Furthermore, the current measurement of the converter or inverter, intended for controlled operation, can be used according to the invention. Thus, no additional hardware expenditure is necessary to implement the invention.

[0023] In a further development, only a single decimation filter can be used, to which the current waveform multiplied by the sinusoidal or cosinusoidal signal is fed several times in succession to determine the respective spectral component at the respective first frequency or k times the respective first frequency. For k = 0, even simple averaging is obtained, so that only a single decimation filter is sufficient overall. The disadvantage of this is that the discrete data stream of the moving average on the output side is only available at long time intervals.

[0024] In an advantageous embodiment, at least one local minimum of the determined frequency-dependent winding impedance, in particular a local minimum of the magnitude of the determined frequency-dependent winding impedance, is determined and monitored for exceeding an unacceptably high degree of deviation from a predetermined value. The advantage here is that condition monitoring of the electric motor is possible in a simple manner and without the need for additional hardware components. This is because the sigma-delta modulator and the decimation filters are already present in a microcontroller of the inverter or converter. Important features of the method for monitoring an electric motor by determining and evaluating the frequency-dependent winding impedance of a three-phase motor fed by an inverter or converter having a current sensor, in particular comprising the aforementioned method, are that:in particular instead of during commissioning or in addition to commissioning, when the rotor shaft of the motor is at a standstill or at a low speed of the rotor shaft of the motor, in particular as condition monitoring, in the controlled operation of the motor fed by the pulse-width modulated converter while a zero vector is present, the winding impedance Z is determined at least at a single first frequency and, in particular, at least the magnitude of the winding impedance Z is monitored for an inadmissibly high degree of deviation from a predetermined threshold value, wherein, to determine the winding impedance Z at the single first frequency, a voltage, in particular a rectangular voltage, the frequency of which is this first frequency, is applied to at least one of the motor phases, wherein the values ​​of the motor current detected by the current sensor, in particular analog, are converted by the sigma-delta modulator into a digital data stream, which is fed to the first decimation filter,the first multiplier, which performs the multiplication with a time-dependent sinusoidal signal whose frequency is this first frequency f, n and the second multiplier, which carries out the multiplication with a time-dependent cosinusoidal signal whose frequency is the first frequency, wherein the output signal of the first multiplier is fed to the second decimation filter for determining a first spectral component, wherein the output signal of the second multiplier is fed to a third decimation filter for determining a second spectral component, wherein the winding impedance Z is determined from these spectral components of the current and from the amplitude of the voltage, in particular by quotient formation, in particular according to in particular where U(f n ) the spectral component of the voltage at the first frequency f n is and l(f n) the Fourier component of the current at the first frequency f n is.

[0025] The advantage here is that condition monitoring is possible during operation, as the impedance can be determined whenever the speed is very low or disappears, in particular at least at a single first frequency, and thus changes can be detected and a warning can be displayed or forwarded depending on the result of the monitoring.

[0026] In an advantageous embodiment, each of the decimation filters has a Sinc 2The output signal of the filter is fed to a sine filter, i.e., a first-order sine filter with a second decimation rate M1, followed by downsampling. The advantage here is that the second-order sine filter suppresses or at least reduces the quantization noise, and the first-order sine filter determines a moving average of its input signal. Thus, sine filters can be used without additional effort because they are integrated into the microcontroller or an FPGA in the converter's signal electronics.

[0027] In an advantageous embodiment, the first decimation rate M1 has a value between 31 and 65. This has the advantage that the quantization noise is suppressed or at least reduced.

[0028] In an advantageous embodiment, the second decimation rate M2 is set such that the first-order sine filter, in particular including downstream downsampling, acts as an averaging element for exactly an integer number of, in particular exactly four, pulse-width modulation periods of the pulse-width modulated power semiconductor switches of the inverter or converter. This is advantageous because averaging occurs over a pulse-width modulation period, thus allowing a moving average to be determined.

[0029] In an advantageous embodiment, the spectral component of the voltage used to determine the winding impedance is taken from the Fourier series of the voltage, in particular the rectangular one. This is advantageous because the spectral component of the voltage is known in advance, since the spectral component of the applied voltage is mathematically given and therefore does not need to be determined mathematically or by measurement.

[0030] In an advantageous embodiment, the respective first spectral components are the imaginary components, and the respective second spectral components are the real components of the current. This has the advantage that the Fourier transform of the current can be easily determined by mixing the sine and cosine signals and integrating them using the first-order sine filter.

[0031] Important features of the drive, comprising an electric motor fed by a converter or inverter, are that the sigma-delta modulator as well as the digital filters and / or sine filters are arranged as on-chip sine filters in a microcontroller or are set up in an FPGA of the inverter or converter.

[0032] The advantage here is that no additional effort is required to implement the method according to the invention; instead, a state-of-the-art inverter or converter can be used. Essentially, therefore, only software adjustments are necessary.

[0033] Important features of the method for determining the reliability of a drive are that the drive has a three-phase motor fed by a converter or inverter, wherein the frequency-dependent winding impedance of the three-phase motor is determined according to a method according to one of the preceding claims, wherein at least one local minimum of the frequency-dependent winding impedance is determined and monitored for exceeding a permissible degree of deviation from a predetermined value. The advantage here is that by determining the frequency-dependent winding impedance, the determination of the local minimum is made possible, and thus it is possible to assess whether anti-resonances occur. Further advantages arise from the subclaims. The invention is not limited to the combination of features of the claims.For the person skilled in the art, further reasonable combination possibilities of claims and / or individual claim features and / or features of the description and / or the figures will arise, in particular from the task and / or the problem arising from a comparison with the prior art.

[0034] The invention will now be explained in more detail using schematic illustrations:

[0035] Figure 1 schematically shows a portion of the method according to the invention, wherein a moving average of the current and its spectral components are determined.

[0036] Figure 2 shows a schematic diagram of an analog-to-digital converter with a sigma-delta modulator 20 and a decimation filter 21.

[0037] In Figure 3, a Sinc K -Decimation filter, comprising a digital low-pass with down-sampling 31, schematically sketched.

[0038] Figure 4 shows a schematic diagram of a decimation filter structure with averaging over a PWM period

[0039] As shown in the figures, in the method according to the invention the analogue current value detected by a current sensor is fed to a sigma-delta converter 4, the output data stream of which is designed as a digital 1-bit data stream and is fed on the one hand to a first decimation filter 1 and on the other hand via multipliers (5, 6) to further decimation filters (2, 3) in order to determine the real part and imaginary part of Fourier components of the current.

[0040] Within the decimation filters (1, 2, 3) the 1-bit data stream is converted by a Sinc 2 -filter into a parallel output data stream with reduced quantization noise.

[0041] From the Fourier components of the current, the current can be represented as a complex quantity and thus the winding impedance can be calculated according to at the voltage U provided by the inverter or converter, the voltage can be determined and monitored for, for example, anti-resonances, local minima, resonance peaks or local maxima in order to increase operational reliability.

[0042] The spectral component of the voltage U at frequency fn is used as U(fn), where n numbers the points in the discretely determined Fourier spectrum. The drive comprises a three-phase motor fed by an inverter or converter via a connecting cable, with the motor current being measured by a current sensor. The current flowing in each phase of the motor is measured or determined. The semiconductor switches of the output stage of the inverter or converter are controlled by pulse-width modulation.

[0043] The output stage has three series circuits connected in parallel, each of the three series circuits having two semiconductor switches, in particular MOSFET or IGBT, connected in series, each of the semiconductor switches being connected in parallel to a freewheeling diode.

[0044] Using pulse-width modulation (PWM) to control the semiconductor switches, the frequency and amplitude of the motor voltage can be freely adjusted by adjusting the pulse-width modulation (PWM) ratio. The PWM switching frequency, for example, is fPWM = 8 kHz, and the fundamental frequency relevant to the motor is 50 Hz. The inductance of the motor winding (LW) smooths the current. Only the fundamental frequency amplitude generates the desired motor torque.

[0045] The inverter or converter is operated in current control mode, and for this purpose, the current is sampled at twice the switching frequency, so that the switching times of each positive and negative PWM switching edge can be used for control. With a switching frequency of 8 kHz, the control loop sampling frequency is fs = 16 kHz. Taking the sampling theorem into account, frequencies up to a maximum of half the sampling frequency – here fs / 2 = 8 kHz – can be analyzed with such a sampled current signal. However, according to the invention, Sinc K -Filter is used and thus spectral components of higher frequencies are also accessible and can be evaluated.

[0046] Antiresonances in the high-frequency behavior of the motor's stator winding indicate unintended voltage stresses that have occurred during operation. Such voltage stresses are the primary cause of premature failure of the winding insulation.

[0047] Minima of the impedance curve at the frequencies fm - so-called anti-resonances - indicate series resonances. In a series circuit, the = capacitive reactance of a parasitic capacitance C m w c m the inductive reactance of a

[0048] Partial winding . Characteristic of a series resonance is that the

[0049] Voltages at the inductance Lm or the capacitance Cm can be significantly higher than the voltage applied to the motor terminals.

[0050] By determining the frequency-dependent winding impedance and its local minima, the load, especially the load accumulated over the operating time, can be monitored for exceeding a permissible deviation from a specified value, thus improving operational reliability. Warning information can be displayed and / or transmitted even before the motor fails.

[0051] To determine the frequency-dependent winding impedance, the

[0052] Drive a first of the phases of the motor with a rectangular symmetrical

[0053] alternating voltage, the Fourier analysis of which can be represented as follows: where fPWM is the switching frequency, for example fPWM = 8 kHz, and

[0054] TPWM is the pulse width modulation duration, for example at a duty cycle of 50%.

[0055] The exemplary switching frequency of 8 kHz of the rectangular voltage therefore has spectral components at f1 = 8 kHz, f3 = 24 kHz, f5 = 40 kHz, f7 = 56 kHz, f9 = 72 kHz, etc.

[0056] For the higher-frequency spectral components, any transistor locking times and their edge steepness must also be taken into account. However, especially when using SiC transistors, which are considered particularly critical with regard to the winding, locking times and rise times are comparatively short.

[0057] By means of the application, a current is driven through the first motor phase winding, the spectral components of which are determined successively for different frequencies by initially selecting k = 1 in the method shown in Figure 1 and thereby feeding the current value multiplied by sin (kwt) to the second decimation filter 2 to determine the imaginary part of the current, i.e. the spectral component T o , in particular where n = k.

[0058] Likewise, the current value multiplied by cos (kwt) is fed to the third decimation filter 3 to determine the real part of the current, i.e. the spectral component

[0059] T ö , in particular where n = k.

[0060] In order to determine the spectral components for further values ​​of k in the same way, the current curve i(t) used for the determination of the spectral components at k=1 is saved and reused accordingly when determining the further values ​​of k.

[0061] The current composed of the spectral components determined in this way can be represented as

[0062] By means of the method according to the invention, higher-frequency spectral components can be determined, in particular spectral components above 8 kHz.

[0063] In the next method step according to the invention, the angular frequency w of the rectangular voltage curve provided by the inverter or converter is changed and, in a corresponding manner, for each such angular frequency w, the spectral components for several values ​​of k are determined.

[0064] By providing a sufficiently large number of rectangular voltage waveforms, whose angular frequencies w differ, provided sequentially by the inverter or converter, and by a sufficiently high number of values ​​of k, a sufficient Fourier spectrum of the current can be obtained so that the local minima in the frequency spectrum can be determined with sufficient precision, thus detecting an unacceptably high deviation from a specified value, and thus monitoring the motor. For example, the two numbers are selected such that the Fourier spectrum can be represented by 100 to 1000 points, in particular, whereby the spectral components for 100 to 1000 frequency values ​​are determined. In particular, such a determination requires between 1 minute and 20 minutes in practice.

[0065] As shown in Figure 2, the detected, analog, continuous-time motor phase currents are converted into a digital bit stream, in particular a 1-bit data stream, using a sigma-delta modulator 20, wherein the sigma-delta modulation frequency is preferably f = 20 MHz. A downstream decimation filter 21 reduces the quantization noise and forms cyclically sampled data words from the motor phase current, in particular a parallel digital data stream.

[0066] As shown in Figure 3, the decimation filter 21 comprises a digital filter 30, in particular Sinc K-Filter, which is configurable via two parameters, namely via the order K, in particular where preferably K = 2 or K = 3, and via the decimation rate M, in particular where preferably M = 128. Higher values ​​for K and M reduce the cutoff frequency of the filter, but improve the suppression of quantization noise and thus the effective resolution of the analog-to-digital conversion.

[0067] Preferably, the sigma-delta modulator 20 is implemented in a second-order integrated circuit and together with integrated galvanic isolation.

[0068] As shown in Figure 3, the digital filter 30, in particular Sinc K -Filter, a downsampling 31 , in particular of order M, can be connected downstream.

[0069] According to the invention, however, the structures shown in Figure 4 are used as the first decimation filter 1, the second decimation filter 2 and the third decimation filter 3.

[0070] As shown in Figure 4, this structure comprises a first digital filter 40, in particular in a first stage of the decimation filter structure, followed by a first downsampling 41, followed by a second digital filter 42, in particular in a second stage of the decimation filter structure, followed by a second downsampling 43.

[0071] The first digital filter is called Sinc 2 -filter, i.e. in second order K = 2. The decimation rate M1 of the first digital filter 40 is a value from the interval 32 to 64. The second digital filter acts as a mean value filter and is called Sinc 1 -filter is executed in first order, K = 1 , and in second order, K = 1 .

[0072] The decimation rate M2 of the second digital filter 42 is selected such that the averaging occurs for exactly an integer number of PWM periods T, i.e., TI = N ■ TPWM. This corresponds to a good approximation of integrating a time profile over its period T, in particular, as in Fourier analysis. The second digital filter 42 thus forms a moving average over exactly one PWM period, synchronously with the PWM.

[0073] The structure shown in Figure 4 is preferably used both as the first decimation filter 1 and as the second decimation filter 2 and also as the third decimation filter 3 in the embodiment according to Figure 1. All three decimation filters (1, 2, 3) are therefore constructed in the same way.

[0074] The first decimation filter 1 is thus used to determine a moving average of the detected current signal.

[0075] By means of the second decimation filter 2, a spectral component, in particular the imaginary part of the Fourier component of the current, is thus determined, in particular since the detected current multiplied by the sine is used as the input signal of the second decimation filter 2.

[0076] By means of the third decimation filter 3, a further spectral component, in particular the real part of the Fourier component of the current, is thus determined, in particular since the detected current multiplied by the cosine is used as the input signal of the third decimation filter 3.

[0077] Preferably, M2 is chosen such that the average is calculated over several complete PWM periods, in particular exactly four PWM periods, i.e., TI = 4 PWM. This allows for improved accuracy despite the stronger quantization noise of the sigma-delta modulators at higher frequencies.

[0078] According to the invention, all intended spectral components of the current are measured one after the other by a targeted variation of the PWM period TPWM or the PWM frequency fpwM. In further embodiments according to the invention, the spectral components are also measured in the manner shown in Figure 1 during controlled operation of the motor when the rotor shaft is stationary or when the rotor shaft of the motor is rotating very slowly while zero vectors are applied in pulse-width modulated operation. Although a densely populated Fourier spectrum cannot be determined, the spectral components can be determined for different values ​​of k at at least one frequency of the rectangular voltage provided by the inverter or converter, and thus the winding impedance of the motor can be monitored at least for some points in the Fourier spectrum.

[0079] In further embodiments according to the invention, the first decimation filter 1 is omitted by also considering the value k=0 in the third decimation filter 3, thus performing the moving average calculation without a cosine-shaped function. In this way, only two decimation filters (2, 3) are necessary.

[0080] In further embodiments according to the invention, each of the decimation filters (1, 2, 3) has, instead of the said sine 2 - and Sinc 1 -Filter with down-sampling each a Sinc 3 -filter, i.e. a third-order sine filter with a first decimation rate M, with down-sampling, where the sinc 3 -filter is triggered synchronously with the pulse width modulation frequency, in particular where M= 128 or 256.

[0081] List of reference symbols

[0082] 1 first decimation filter

[0083] 2 first decimation filter

[0084] 3 first decimation filter

[0085] 4 Current sensor with sigma-delta modulator

[0086] 5 multipliers

[0087] 6 multipliers

[0088] 20 Sigma-Delta Modulator

[0089] 21 decimation filters

[0090] 30 digital filters, especially Sinc K -Filter

[0091] 31 Downsampling

[0092] 40 first digital filter, especially in the first stage of the decimation filter structure

[0093] 41 Downsampling

[0094] 42 second digital filter, especially in the second stage of the decimation filter structure

[0095] 43 Downsampling

[0096] K Ordinal number of the harmonic w Angular frequency

[0097] T integration variable

[0098] T Period t Time i Current f n frequency

Claims

Patent claims:

1. Method for determining the frequency-dependent winding impedance of a three-phase motor fed by an inverter or converter having a current sensor, characterized in that, in particular during commissioning of the three-phase motor, successively in time intervals for different first frequencies f n in each case a time-dependent, in particular rectangular, voltage waveform, the frequency of which is the respective first frequency f nis applied to at least one of the motor phase windings, wherein the values ​​of the motor current detected by a current sensor, in particular analog, are converted by a sigma-delta modulator into a digital data stream, in particular into a 1-bit data stream, which is fed to a first decimation filter to form a moving average, in particular which is used as a spectral component at vanishing frequency, is multiplied for several natural numbers k in each case by a time-dependent sinusoidal signal, the frequency of which is the respective k-fold of the respective first frequency f n and a moving average of the resulting product is formed to determine a respective first spectral component, in particular imaginary component, and for the natural numbers k is multiplied by a time-dependent cosinusoidal signal whose frequency is the respective k-fold of the respective first frequency f nand a moving average of the resulting product is formed to determine a respective second spectral component, in particular real component, wherein the winding impedance Z is determined from the spectral components of the current and the voltage for each of the first frequencies and their k-fold multiples, in particular by quotient formation, in particular according to in particular where U(f n ) the spectral component of the voltage at the first frequency f n is and l(f n ) the Fourier component of the current at the first frequency f n is.

2. Method for determining the frequency-dependent winding impedance of a three-phase motor fed by an inverter or converter having a current sensor, characterized in that, in particular during commissioning of the three-phase motor, successively in time intervals for different first frequencies f nin each case a time-dependent, in particular rectangular, voltage waveform, the frequency of which is the respective first frequency f n is applied to at least one of the motor phases, wherein the values ​​of the motor current detected by a current sensor, in particular analog, are converted by a sigma-delta modulator into a digital data stream, in particular into a 1-bit data stream, which is fed to a first decimation filter, a first multiplier, which for a plurality of natural numbers k each carries out a multiplication with a time-dependent sinusoidal signal, the frequency of which is the respective k-fold of the respective first frequency f n and a second multiplier which carries out a multiplication for each of the plurality of natural numbers k with a time-dependent cosinusoidal signal whose frequency is the respective k-fold of the respective first frequency f nis supplied, wherein the output signal of the first multiplier is supplied to a second decimation filter for determining a respective first spectral component, in particular imaginary component, wherein the output signal of the second multiplier is fed to a third decimation filter for determining a respective second spectral component, in particular a real component, wherein the winding impedance Z is determined from the spectral components of the current and the voltage for each of the first frequencies and their K-fold frequencies, in particular by quotient formation, in particular according to in particular where U(f n ) the spectral component of the voltage at the first frequency f n is and I (f n ) the Fourier component of the current at the first frequency f nis, in particular wherein the respective spectral component of the voltage is determined from the voltage curve applied by the inverter or converter, in particular mathematically.

3. Method according to one of the preceding claims, characterized in that at least one local minimum of the determined frequency-dependent winding impedance, in particular a local minimum of the magnitude of the determined frequency-dependent winding impedance, is determined and monitored for exceeding an inadmissibly high degree of deviation from a predetermined value.

4. A method for monitoring an electric motor by determining and evaluating the frequency-dependent winding impedance of a three-phase motor fed by an inverter or converter having a current sensor, in particular comprising a method according to claim 1 or 2, characterized in that, in particular instead of during commissioning or in addition to commissioning, when the rotor shaft of the motor is at a standstill or at a low speed of the rotor shaft of the motor, in particular as condition monitoring, during the controlled operation of the motor fed by the pulse-width-modulated converter, the winding impedance Z is determined during the application of a zero vector at least at a single first frequency and, in particular, at least the magnitude of the winding impedance Z is monitored for an inadmissibly high degree of deviation from a predetermined threshold value,wherein, to determine the winding impedance Z at the single first frequency, a particularly rectangular voltage, the frequency of which is this first frequency, is applied to at least one of the motor phases, wherein the values ​​of the motor current detected by the current sensor, in particular analog, are converted by the sigma-delta modulator into a digital data stream, in particular into a 1-bit data stream, which is passed to the first decimation filter, the first multiplier, which carries out the multiplication by a time-dependent sinusoidal signal, the frequency of which is this first frequency f, n and the second multiplier, which carries out the multiplication with a time-dependent cosinusoidal signal whose frequency is the first frequency, wherein the output signal of the first multiplier is fed to the second decimation filter for determining a first spectral component, in particular imaginary part, wherein the output signal of the second multiplier is fed to a third decimation filter for determining a second spectral component, in particular real part, wherein the winding impedance Z is determined from these spectral components of the current and from the amplitude of the voltage, in particular by quotient formation, in particular according to in particular where U(f n ) the spectral component of the voltage at the first Frequency f n is and l(f n ) the Fourier component of the current at the first frequency f n is.

5. Method according to one of the preceding claims, characterized in that each of the decimation filters comprises a sinc 2-filter, i.e. a second-order sine filter, with a first decimation rate M1, with down-sampling, whose output signal is fed to a sine filter, i.e. a first-order sine filter with a second decimation rate M2, with down-sampling, or that each of the decimation filters has a sinc 3 -filter, i.e. a third-order sine filter with a first decimation rate M, with down-sampling, where the sinc 3 - Filter is triggered synchronously to the pulse width modulation frequency, in particular where M= 128 or 256.

6. Method according to one of the preceding claims, characterized in that the first decimation rate M1 has a value between 31 and 65.

7. Method according to one of the preceding claims, characterized in that the second decimation rate M2 is set such that the first-order sine filter, in particular including downstream downsampling, is effective as an averaging for exactly an integer number of, in particular for exactly four, pulse width modulation period durations of the pulse width modulated controlled power semiconductor switches of the inverter or converter.

8. Method according to one of the preceding claims, characterized in that the spectral component of the voltage used to determine the winding impedance is taken from the Fourier series of the particularly rectangular voltage.

9. Method according to one of the preceding claims, characterized in that the respective first spectral components are the imaginary components and the respective second spectral components are the real components of the current.

10. Method according to one of the preceding claims, characterized in that the first multiplier and the second multiplier are arranged in an FPGA of the converter or inverter, wherein the digital data stream, in particular 1-bit data stream, is also supplied to the FPGA.

11. Drive, comprising an electric motor fed by a converter or inverter, for carrying out a method according to one of the preceding claims, characterized in that the sigma-delta modulator and the respective decimation filter, in particular the digital filters, and / or sine filters are arranged as on-chip sine filters in a microcontroller or are set up in an FPGA of the inverter or converter.

12. Drive according to one of the preceding claims, characterized in that the data stream emerging from the output side of the sigma-delta modulator, in particular a 1-bit data stream, is fed to the respective multiplier for multiplication by the sinusoidal or cosinusoidal signal.

13. Method for determining the reliability of a drive, in particular according to claim 11 or 12, wherein the drive comprises a three-phase motor fed by a converter or inverter, wherein the frequency-dependent winding impedance of the three-phase motor is determined according to a method according to one of the preceding claims, characterized in that at least one local minimum of the frequency-dependent winding impedance is determined and monitored for exceeding a permissible degree of deviation from a predetermined value.

14. Drive for carrying out a method according to claim 13, characterized in that the drive comprises a three-phase motor fed by a converter or inverter.

Images