Method for Detecting Abnormality of Motor During Transition Operation

Abstract

The present invention relates to a method and a device for detecting an abnormality of a motor during a transient operation of the motor. The present invention includes detecting at least one motor signal, identifying a fundamental frequency of the motor signal, where the fundamental frequency is the frequency of a voltage waveform that drives a stator of the motor, performing heterodyne non-coherent demodulation of the motor signal in a frequency band, where a center frequency of the frequency band and a bandwidth of the frequency band have a fixed proportional ratio with the identified fundamental frequency, and determining a level of an abnormality in the motor by comparing the demodulated signal with a signal derived from the motor signal.

Description

Technical Field

[0001] The present invention generally relates to a method for detecting motor abnormalities during transient operation.

Background Art

[0002] In a rotating electrical machine, for example, when the rotation axes of the stator and rotor are misaligned, when the bearing is damaged, when the rotor bar in the stator is broken, or when the insulation of the winding is damaged, etc., abnormalities in the motor may occur. Such abnormalities can cause wear of the motor bearing, unexpected vibration of the generated torque, and undesirable vibrations, which can damage the plant.

[0003] In the prior art, a vibration sensor or a current sensor, or a combination of both, can be used to detect motor abnormalities. Each motor abnormality has a specific frequency signature, which can be detected from the measured current waveform. MCSA (Motor Current Signal Analysis) technology is widespread because it reuses the current sensor required for closed-loop control of the motor. The level and frequency of the abnormal harmonics vary depending on the rotational speed, the applied torque, and the type and level of the abnormality.

[0004] In an asynchronous machine, the rotor frequency is different from the stator frequency, and the difference (rotor slip) is generated by the application of torque. Since many applications are not equipped with a position sensor, it is difficult to estimate the rotor frequency and / or the rotor slip.

[0005] The popular MCSA method estimates the presence of motor abnormalities using time-frequency conversion (e.g., Fourier transform, wavelet transform, etc.) of the input (e.g., phase current) waveform. In the Fourier method, the input waveform is decomposed at the fundamental part of the equally spaced gyrators. In the SFFT method or STFT method, the time-varying changes of the gyrator components can be tracked in both steady-state applications and transient applications.

[0006] There are many problems in the prior art. In the FFT method, it is necessary to collect, store, and process waveforms for a long time (e.g., 10 seconds in the case of 0.1 Hz accuracy). Therefore, this process is hardly suitable for the implementation of motor controllers.

[0007] The measured waveform contains strong components at the stator frequency. This basic signal is superimposed on the abnormal signature signal and is several orders of magnitude larger than the abnormal signature signal.

[0008] The abnormal frequency varies with the rotor frequency but is generally unknown due to rotor slip. The frequency of the abnormal signature signal can be estimated using an FLL (Frequency Lock Loop), but the FLL is more likely to lock in the strong basic signal rather than the abnormal signature signal, and lock errors tend to occur especially in the case of noisy signals. In a machine, two or more types of abnormalities may occur.

[0009] Such techniques cannot detect motor abnormalities during the transient operation of the machine. In many applications, the operating conditions (speed and torque) are not static and cannot be maintained over a long period. SUMMARY OF THE INVENTION

[0010] The present invention is a method for detecting motor abnormalities during the transient operation of a motor, comprising: detecting at least one motor signal; A step of identifying the fundamental frequency of a motor signal, where the fundamental frequency is the frequency of the voltage waveform driving the stator of the motor, and the step; A step of performing heterodyne non-coherent demodulation of the motor signal in a frequency band, where the center frequency of the frequency band and the bandwidth of the frequency band have a fixed proportional ratio with the identified fundamental frequency, and the step; A step of determining the level of abnormality in the motor by comparing the demodulated signal with a signal derived from the motor signal; The object is to provide a method characterized by including the above.

[0011] The present invention is a device for detecting an abnormality of a motor during a transient operation of the motor, Means for detecting at least one motor signal; Means for identifying the fundamental frequency of the motor signal, where the fundamental frequency is the frequency of the voltage waveform driving the stator of the motor; Means for performing heterodyne non-coherent demodulation of the motor signal in a frequency band, where the center frequency of the frequency band and the bandwidth of the frequency band have a fixed proportional ratio with the identified fundamental frequency; Means for determining the level of abnormality in the motor by comparing the demodulated signal with a signal derived from the motor signal; The present invention also relates to a device characterized by comprising the above.

[0012] Therefore, as long as the frequency of the abnormal signature signal is included in the above frequency band, the level of abnormality can be estimated. The level of abnormality (such as the eccentricity level in the case of an eccentricity abnormality) can be obtained from the level of the detected abnormal signature signal. The level of the abnormal signature signal is estimated even when the exact frequency of the abnormal signature signal is not fully known, such as when there is a slip between the stator frequency and the rotor frequency of an induction machine.

[0013] The level of the abnormal signature signal can be estimated without performing complex time-frequency signal analysis such as SFFT. Detection of motor abnormalities can be achieved with low complexity and can be carried out with a general-purpose inverter without adding sensors and / or computing capabilities.

[0014] The level of the abnormality is estimated in real time for continuous signal inputs. In this method, there is no need to store data in a large analysis window. The level of the abnormality is estimated in a short time and is robust to sudden changes in the operating state (torque and speed) of the machine.

[0015] Since the frequency of the motor abnormality is always obtained as a weighted sum of the stator frequency and the rotor frequency, the processing is simple and effective for all rotor speeds. The ratio of the bandwidth to the stator frequency can be selected such that the abnormal signature signal is always maintained within the selected frequency band due to the change in the rotor frequency due to slip. Since the selected frequency band is indexed to the stator frequency, this method can track the level of the abnormality under transient speed conditions.

[0016] This method is effective in applications that hardly or do not satisfy a long steady state at all.

[0017] According to a specific feature, non-coherent demodulation includes mixing the motor signal with a sine wave and a cosine wave oscillating at the center frequency, low-pass filtering the mixed signal using a cut-off frequency equal to the bandwidth of the frequency band, combining the low-pass filtered signals, and.

[0018] Therefore, as long as the frequency of the abnormal signature signal belongs to the above frequency band, that is, as long as the distance from the frequency of the abnormal signature signal to the center frequency is smaller than the distance to the cut-off frequency, it can be demodulated using another reference signal located at the center frequency. The exact frequency of the abnormal signature signal does not have to be estimated. In this method, a systematic search over all frequencies such as the time-frequency analysis method is not necessary. In this method, the estimation of the rotational frequency is also not required, so the risks and costs associated with the estimation or detection of the rotor position are reduced.

[0019] Since energy is collected in both the in-phase component and the quadrature component of the reference signal, the level of the abnormal signature signal is estimated without energy loss or unwanted vibrations.

[0020] Interference signals such as second harmonics caused by the mixer are suppressed as long as their frequencies are away from the above frequency band. By strictly setting the fixed ratio, all unwanted harmonics that may interfere with the estimation of abnormalities can be removed. Different types of abnormalities are located in different segments of the spectrum, so it is possible to distinguish different types of abnormalities.

[0021] According to a specific feature, the method is a step performed before non-coherent demodulation, a step of notch filtering the motor signal, wherein the center frequency of the notch filter is equal to the specified fundamental frequency, the width of the notch filter has a fixed proportional ratio with the fundamental frequency, and the center frequency of the frequency band is outside the band of the notch filter, further includes.

[0022] Therefore, demodulation in the above frequency band is robust against the detection of minute abnormal signature signals when the above frequency band is adjacent to a strong current signal flowing at the fundamental frequency of the motor. Cross-channel interference is minimized, and in this method, minute abnormal levels can be detected. This method is suitable for motor condition monitoring. In this state, the condition of the machine is regularly monitored from the initial zero-abnormal state to the severe abnormal state, and predictive maintenance can be planned and abnormalities repaired before reaching the severe abnormal state, so there is no risk of suddenly stopping the application in case an accident caused by a severe abnormality occurs.

[0023] According to a specific feature, the center frequency of the frequency band, the bandwidth of the frequency band, and the proportional ratio between the center frequency of the notch filter and the fundamental frequency are determined in advance for a given type of abnormality.

[0024] Therefore, by simply repeating this method using various sets of proportional ratios, several types of abnormalities can be monitored at once. The abnormal signature exhibits harmonics at positions that combine an integer number of stator frequencies and an integer number of rotor frequencies. Although the rotor frequency is generally unknown, it differs from the stator frequency only by the amount of frequency slip, so an upper limit can be set. In fact, the maximum slip is related to the maximum torque of the motor. As a result, the frequency range where abnormal harmonics exist can be accurately determined.

[0025] According to a specific feature, this method is a step performed before heterodyne non-coherent demodulation and / or notch filtering, a step of angle resampling the motor signal, wherein the sampled signal has an equal-phase increment of a signal rotating at a specified fundamental frequency, and further includes.

[0026] Accordingly, the number of signal samples to be processed is reduced, and the numerical complexity of the method is reduced. Since the center frequency and bandwidth of the frequency band have a fixed proportional ratio to the fundamental frequency and the signal is resampled using equal phase increments of the signal rotating at the fundamental frequency, subsequent processing for detecting motor anomalies becomes independent of the fundamental frequency and is robust to changes in the stator frequency. Therefore, the present method is suitable for tracking the abnormal level during the transient state.

[0027] According to a particular feature, notch filtering and low-pass filtering are implemented as IIR filters with fixed predetermined coefficients.

[0028] Accordingly, the complexity in the implementation of the notch filter and the low-pass filter (LPF) becomes very low. In IIR filtering, the memory of past motor signal samples is minimal (usually only a maximum of 4 samples, much less compared to the SFFT method which requires thousands of samples).

[0029] The motor signal is resampled using equal phase increments, and since the bandwidth and cut-off frequency of the filter are proportional to the speed level, the implementation of the IIR filter is realized using constant internal coefficients even when the fundamental frequency changes during a transient state, such as during a speed ramp.

[0030] The coefficients of the notch filter and the low-pass filter do not need to be adapted to changing speed conditions. A rapid change in the IIR coefficients may introduce instability in the output generated by the filter, which is contrary to the purpose of tracking abnormal conditions with high accuracy even when the operating conditions change rapidly. In contrast, the present method of implementing IIR using fixed coefficients is optimized to track the level of the abnormal signature of a motor whose speed is in a transient state. According to a particular feature, the present method performing envelope detection on the detected motor signal; a step of obtaining a proportional ratio between an abnormal harmonic level and an abnormal level in a motor from an envelope line and a fundamental frequency; a step of performing angular resampling of the proportional ratio; a step of notch-filtering the resampled proportional ratio; a step of low-pass filtering the resampled and notch-filtered proportional ratio; further comprising.

[0031] Accordingly, the abnormal level is obtained from the level of the demodulated abnormal signature signal. The application user can know how to set the timing for triggering preventive maintenance to repair the abnormal level based on the change in the abnormal level. As an example, in the case of an eccentricity abnormality, it may gradually increase and the minimum distance between the rotor and the stator may decrease. Before the eccentricity becomes excessively large and the rotor collides with the stator, which may result in catastrophic consequences for the mechatronics chain in some cases, technicians can be dispatched to the site to correct the eccentricity.

[0032] Since the proportional ratio passes through exactly the same signal processing chain (notch filter + low-pass filter), the same delay occurs from this processing chain, and the final result becomes more robust to changes in the signal envelope line (torque) or the fundamental frequency. The estimation of the abnormal level is robust to fast transient states.

[0033] The proportional ratio compensates for changes in abnormal harmonics associated with the torque level and the speed level.

[0034] The features of the present invention will become clearer by reading the following description of the exemplary embodiments. This description is created with reference to the accompanying drawings.

Brief Description of the Drawings

[0035]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7A

Figure 7B

DETAILED DESCRIPTION OF THE INVENTION

[0036] FIG. 1 shows a first example of a device for detecting an abnormality of a motor during a transient operation according to the present invention.

[0037] A device for detecting an abnormality of a motor during a transient operation includes means 103 for specifying a fundamental frequency f s of a motor signal, means for performing heterodyne non-coherent demodulation of the motor signal in a frequency band, and means for obtaining a level of an abnormality in the motor by comparing a demodulated signal with a signal derived from the motor signal. Here, the fundamental frequency is the frequency of a voltage waveform for driving a stator of the motor. Also, a center frequency and a bandwidth of the frequency band are proportional to the specified fundamental frequency.

[0038] The motor signal is, for example, the waveform of the current flowing through the stator winding of the motor. In another example, the device includes an inverter and a controller, and the motor signal is a reference voltage that is obtained by the controller and used for controlling the inverter that supplies power to the motor. In yet another example, the device includes a vibration sensor, and the motor signal is a vibration signal detected by the vibration sensor.

[0039] The device for detecting an abnormality of the motor during a transient operation may further include an angular resampling unit of the motor signal and an infinite impulse response filter. The device for detecting an abnormality of the motor during a transient operation may further include a notch filter.

[0040] The device for detecting an abnormality of the motor during a transient operation may further include a scoring function unit. This scoring function unit scales the level of the abnormal harmonic detected by heterodyne non-coherent demodulation to the level of the abnormality of the motor using a ratio adapted to the transient speed and current state received by the motor.

[0041] The scoring function unit includes an envelope detection module 101 that obtains the norm ||x(kT)|| of the motor signal x(kT). The norm ||x(kT)|| is provided to the abnormal harmonic model module 102. The abnormal harmonic model module 102 obtains a proportional ratio R(kT) between the abnormal harmonic level and the level of the abnormality in the motor.

[0042] The scoring function unit may further include an angular resampling module 108 of the motor signal, a notch filter 110, and an infinite impulse response filter 113.

[0043] The basic detection module 103 obtains the angular frequency f s (kT) of the motor signal x(kT).

[0044] The angular frequency f s (kT) is provided to the abnormal harmonic model module 102 and the phase determination module 104.

[0045] The phase determination module 104 will be described in more detail with reference to FIG. 4.

[0046] In the first embodiment of the present invention, the phase determination module 104 provides the downsampling ratio N to the angle resampling modules 105, 106, and 108, and provides the phase signal θ(kT) to the angle resampling module 106.

[0047] In the second embodiment of the present invention, the phase determination module 104 provides the phase signal θ(kT) to the angle resampling modules 105, 106, and 108.

[0048] The concept of angle resampling is to generate a set of samples with equal phase distances from samples with equal time distances. Angle resampling can implement notch filters and low-pass filters as IIR filters with constant coefficients, along with a fixed ratio between the fundamental frequency, filter bandwidth, and filter cut-off frequency, even in situations where the fundamental frequency changes, such as when there is a speed ramp.

[0049] In the first embodiment of the present invention, angle resampling is to downsample the input signal using a ratio N that is inversely proportional to the fundamental frequency, as shown in the following equation.

Equation

[0050] In the second embodiment of the present invention, angle resampling is to select samples of the input signal corresponding to the time points when the obtained phase angle intersects the ladder of levels having an interval of Δθ.

[0051] For example, Δθ is selected to represent a value between 1 degree and 20 degrees.

[0052] The angle resampling module 105 provides the angle sample x(nΔθ) to the notch filter 109.

[0053] The filtered angle sample x’(nΔθ) is provided to the heterodyne non-coherent demodulation module 100.

[0054] The angle resampling module 106 provides the angle sample nΔθ to the heterodyne non-coherent demodulation module 100.

[0055] The heterodyne non-coherent demodulation module 100 will be described in more detail with reference to FIG. 2.

[0056] The angle resampling module 108 provides the angle sample R(nΔθ) to the notch filter 110.

[0057] The output of the notch filter 110 is provided to the low-pass filtering module 113.

[0058] The device for detecting motor abnormalities during transient operation of the motor is the ratio x between the bandwidth of the notch filter and the fundamental frequency n and includes a normalized notch ratio determination module 107 that provides it to the notch filters 109 and 110.

[0059] The device for detecting motor abnormalities during transient operation of the motor is the ratio x between the cut-off frequency of the low-pass filter and the fundamental frequency c and includes a low-pass filter frequency determination module 112 that provides it to the low-pass filter 113 and the heterodyne non-coherent demodulation module 100.

[0060] The device for detecting motor abnormalities during transient operation of the motor is the ratio x between the demodulation frequency and the fundamental frequency m and includes a demodulation frequency determination module 111 that provides it to the heterodyne non-coherent demodulation module 100.

[0061] The set of three ratios (xn , x c , x m ) is selected to be located within the search window f fault / f fundamental [x m -x c / 2; x m +x c / 2]. As an example, if the abnormality is an eccentricity abnormality and x m = 0.5, x c = 0.1, then the associated lower sideband (f s -f r ) abnormal frequency is located within the window f fault / f fundamental [0.4; 0.6], so as long as the frequency slip is less than 10%, the eccentricity abnormality can be detected. The harmonic (f s -f r ) is always located within the selected demodulation band, so demodulation is effective for any frequency slip. Therefore, this method is robust against changes in the rotational speed f r caused by changes in the torque applied to the motor by the load.

[0062] As another example, if x m = 1.5, x c = 0.1, then the associated upper sideband (f s +f r ) abnormal frequency is located within the window f fault / f fundamental [1.4; 1.6], so as long as the frequency slip is less than 10%, the eccentricity abnormality can be detected.

[0063] As yet another example, if x m = 3.5, x c = 0.1, then it is possible to detect the upper sideband abnormal harmonic around the third fundamental harmonic. More generally, the set of three ratios can be determined in advance for any frequency signature associated with each type of motor abnormality, and this method is equally applicable to the detection of any abnormality.

[0064] A device for detecting an abnormality in a motor during a transition operation includes a score determination module 114 that determines an estimated level of the motor abnormality by comparing a demodulated signal from a heterodyne non-coherent demodulation module 100 with a signal from an output of a low-pass filter 113 derived from a signal provided by the motor.

[0065] The level of abnormal harmonics varies with speed and current. A model can relate the level of abnormal harmonic H fault_est to the level of speed f s and the level of current I. As an example, the estimated level of the motor abnormality is given by the formula H fault_est =A*Fault*f s B *I C where A, B, and C are constants.

[0066] As another example, H fault_est =A*Fault(1 + B*f s ) where A and B are constants.

[0067] When there is a high-speed speed ramp, the demodulated signal H fault reaches with a somewhat delay compared to the instantaneous speed and current due to the delay caused by the notch filter and the IIR filter.

[0068] According to the present invention, the model ratio K = H fault_est / Fault is passed through a signal processing chain that mimics the delay induced by the signal processing chain. This includes an angle resampling module 108, a notch filter 110, and a low-pass filter 113. Using the resulting ratio, the output H fault of the heterodyne demodulation is divided to generate an abnormality estimate Fault_est = H fault / K.

[0069] FIG. 2 shows an example of a block diagram of a heterodyne demodulation module used in the present invention.

[0070] The heterodyne non-coherent demodulation module 100 includes three multiplication modules 200, 201, and 203, a cosine determination module 202, a sine determination module 204, two low-pass filters 205 and 206, and an I&Q combiner 207.

[0071] The multiplication module 200 multiplies the demodulation frequency ratio x m by the angular sample nΔθ. The result of the multiplication module 200 is provided to the cosine determination module 202. The output of the cosine determination module 202 is provided to the multiplication module 201. The result of the multiplication module 200 is further provided to the sine determination module 204. The output of the sine determination module 204 is provided to the multiplication module 203.

[0072] The multiplication module 201 is an in-phase mixer that multiplies the filtered angular sample x'(nΔθ) by the output of the cosine determination module 202.

[0073] The output of the multiplication module 201 is provided to the low-pass filter 205.

[0074] The multiplication module 203 is a quadrature mixer that multiplies the filtered angular sample x'(nΔθ) by the output of the sine determination module 204.

[0075] The output of the multiplication module 203 is provided to the low-pass filter 206.

[0076] The low-pass filters 205 and 206 are preferably infinite impulse response filters having a cut-off frequency ratio x c .

[0077] The outputs of the low-pass filters 205 and 206 are respectively from (x m -x c / 2)*f s to (x m +x c / 2)*f sRepresents the in-phase demodulation result and the quadrature demodulation result of the signals existing in the band up to. These outputs are provided to the I&Q combiner 207. The I&Q combiner 207 outputs the square root of the sum of the squares of the in-phase channel value and the quadrature channel value.

[0078] The output of the I&Q combiner 207 is provided to the score determination module 114.

[0079] Bandwidth f of the notch filter n , mixer frequency f m and LPF frequency f c and the stator frequency f s and the respective ratios x n , x m , x c are constant. In the case of a machine having two pole pairs, as an example, x m = 0.5, x c = 1 / 12 and x n are given by 1 / 12.

[0080] Other ratios are possible and they can be selected to detect the frequency components of various abnormal modes such as static / dynamic eccentricity anomalies, rotor bar breakage, etc. Their signature signals are in the band defined by [f m - f c , f m + f c = f s *[x m - x c , x m + x c .

[0081] The level of the abnormal signature is usually small compared to the basic signal. The notch filter first reduces all components within the bandwidth f n around the stator frequency. The mixer moves all frequency components to the left so that the corresponding frequency component f m is located at DC. The LPF has a demodulation bandwidth f cTo attenuate all signals from this point, the remaining fundamental signals and higher harmonics disappear. The I&Q combiner sums the energies of the in-phase and quadrature channels to remove the phase uncertainty of the unknown signal located in the desired band. The I&Q combiner estimates the level of the single-tone harmonic signal whose frequency is included in the frequency band m -f c ,f m +f c .

[0082] FIG. 3 shows an example of a block diagram of an infinite impulse response filter used according to the present invention.

[0083] The IIR filter is implemented using a set of delay lines 300, 301, 302, 316, 317, 318, amplifiers 303, 304, 305, 306, 313, 314, 315, and adders 307, 308, 309, 310, 311, 312. Since its complexity is very small, the IIR filter can be easily implemented either in software or in hardware. The IIR filter can filter the input signal in real time by storing the past signal events to a minimum. The filter coefficient b k is applied to the input, and the coefficient a k is applied to the output.

[0084] The filter coefficients are determined so as to obtain the desired spectral response. The IIR filter can be obtained, for example, from an analog filter transfer function through bilinear approximation, or impulse invariant transformation.

[0085] In the case of a first-order notch filter, the desired response function is

Equation

Equation

[0086] For resampling, F s / ω s is constant, and ω n / ω s is selected to be constant, so the filter coefficients do not need to be recalculated even when the speed changes.

[0087] In the case of a third-order Butterworth low-pass filter, the desired response function is

Equation

Equation

[0088] For resampling, F s / ω s is constant, and x c = ω c / ω s is selected to be constant, so the filter coefficients do not need to be recalculated even when the speed changes.

[0089] In a preferred embodiment, both F s and ω c are selected to be normalized with respect to the stator speed. That is, F s = 1 / Δθ, ω c = f c / f s is. This makes the change small even when the speed changes very rapidly between consecutive samples inside the IIR filter.

[0090] Filters 109, 110, 113, 205, and 206 are infinite impulse response filters.

[0091] The IIR filter shown in FIG. 3 is an m-th order filter.

[0092] As shown in FIG. 3, the input IN of the IIR filter is provided to a multiplication module (or amplifier) 303 that multiplies the input sample by the coefficient b0 and an inverse Z-transform module (or delay line) 300.

[0093] The output of the inverse Z-transform module 300 is provided to a multiplication module 304 that multiplies the sample by the coefficient b1 and an inverse Z-transform module 301.

[0094] The output of the inverse Z-transform module 301 is provided to a multiplication module 305 that multiplies the sample by the coefficient b2 and an m-th inverse Z-transform module 302.

[0095] The output of the inverse Z-transform module 302 is provided to a multiplication module 306 that multiplies the sample by the coefficient b M and is provided to a multiplication module 306 that multiplies the sample by the coefficient b.

[0096] The output of the m-th multiplication module 306 is added to the output of the multiplication module 305 by an addition module (or adder) 309.

[0097] The output of the addition module 309 is added to the output of the multiplication module 304 by an addition module 308.

[0098] The output of the addition module 308 is added to the output of the multiplication module 303 by an addition module 307.

[0099] The output of the addition module 307 is provided to an addition module 310 that provides the filtered output sample OUT.

[0100] The output of the addition module 310 is provided to an inverse Z-transform module 316.

[0101] The output of the inverse Z-transform module 316 is provided to a multiplication module 313 that multiplies the samples by the coefficient -a1 and an inverse Z-transform module 317.

[0102] The output of the inverse Z-transform module 317 is provided to a multiplication module 314 that multiplies the samples by the coefficient -a2 and an m-th inverse Z-transform module 318.

[0103] The output of the inverse Z-transform module 318 is provided to a multiplication module 315 that multiplies the samples by the coefficient -a M to.

[0104] The output of the m-th multiplication module 315 is added to the output of the multiplication module 314 by an addition module 312.

[0105] The output of the addition module 312 is added to the output of the multiplication module 313 by an addition module 311.

[0106] The output of the addition module 311 is added to the output of the addition module 307 by an addition module 310.

[0107] FIG. 4 shows an example of a block diagram of a phase module of a device for detecting an abnormality of a motor in a transient operation according to the present invention.

[0108] The phase determination module 104 includes an amplifier 400, an integrator 401 that derives an angle from the angular frequency ω(kT), and a sampling rate determination module 402 that obtains the downsampling ratio N as s from

Number

[0109] FIG. 5 shows a second example of a device for detecting an abnormality of a motor during a transient operation according to the present invention.

[0110] A device 50 for detecting an abnormality of a motor during a transient operation of the motor has an architecture based on, for example, components connected by a bus 501 and a processor 500 controlled by a program disclosed in FIG. 6.

[0111] The bus 501 links the processor 500 to a read-only memory ROM 502, a random access memory RAM 503, and an input / output interface I / O IF 505.

[0112] By means of the input / output interface I / O IF 505, the device 50 can detect a motor signal that can include a spectral signature of an abnormality of the motor.

[0113] The memory 503 includes registers for accommodating program variables and instructions related to the algorithm disclosed in FIG. 6.

[0114] The read-only memory, or in some cases a flash memory 502, includes program instructions related to the algorithm disclosed in FIG. 6. These instructions are loaded into the random access memory 503 when the device 50 is powered on. Alternatively, the program can also be executed directly from the ROM memory 502.

[0115] The calculations performed by the device 50 can also be implemented in software by executing a series of instructions or programs by a programmable computing machine such as a PC (personal computer), DSP (digital signal processor), or microcontroller, or otherwise can be implemented in hardware by a machine or dedicated components such as an FPGA (field programmable gate array) or ASIC (application specific integrated circuit).

[0116] In other words, the device 50 includes a circuit section that causes the device 50 to execute a program related to the algorithm disclosed in FIG. 6, or a device including the circuit section.

[0117] Figure 6 shows an example of an algorithm for detecting motor anomalies during transient operation according to the present invention.

[0118] This algorithm is disclosed in an example executed by the processor 500 of device 50.

[0119] In the first step S600, the processor 500 determines the ratio x n between the bandwidth f m of the notch filter, the demodulation frequency f c , the bandwidth f n of the demodulation, and the fundamental frequency of the motor signal. These ratios are constant for a given anomaly type. As an example, monitoring the upper sideband harmonics around the first harmonic of an eccentricity anomaly corresponds to (x m , x c ) = (1 / 12, 3 / 2, 1 / 12). The processor 500 then proceeds to step S601. n , x m , x c ) = (1 / 12, 3 / 2, 1 / 12). The processor 500 then proceeds to step S601.

[0120] In step S601, the processor 500 detects the motor signal sent from the I / O interface 505 and proceeds to step S602.

[0121] In step S602, the processor 500 determines the fundamental frequency f s of the motor signal detected in step S601. This fundamental frequency is the frequency of the voltage waveform that drives the stator of the motor.

[0122] The motor signal is, for example, the current waveform flowing through the stator windings of the motor. In another example, the motor signal is the reference voltage required by the controller and used to control the inverter that supplies power to the motor. The motor signal may be a detected vibration signal detected by a vibration sensor.

[0123] In the next step S603, the processor 500 obtains a phase by integrating the fundamental frequency obtained in step S602.

[0124] In the first embodiment of the present invention, by obtaining the phase, the downsampling ratio N and the phase signal θ(kT) of the following formula are obtained.

Equation

[0125] In the second embodiment of the present invention, by obtaining the phase, the phase signal θ(kT) is obtained.

[0126] In step S604, the processor 500 performs angle resampling of the motor signal x(kT) in order to provide a signal sample x(nΔθ) using regular angle sampling.

[0127] In step S605, the processor 500 performs angle resampling of the phase signal that provides a phase sample nΔθ using regular angle sampling.

[0128] In step S606, the processor 500 performs filtering of the resampled signal x(nΔθ) provided in the angle resampling step S604. This filtering is notch filtering disclosed in FIG. 1.

[0129] In step S607, the processor 500 uses the phase sample nΔθ and the ratios x m and x c to perform heterodyne non-coherent demodulation of the filtered resampled signal x'(nΔθ) provided in the filtering step S606. The processor 500 uses the phase sample nΔθ as the ratio x mMultiply and calculate the sine and cosine functions of the multiplied phase. The processor 500 mixes the filtered resampled motor signal with the results of the sine and cosine functions. The processor 500 applies an IIR low-pass filtering step to the acquired in-phase signal and quadrature signal. The processor 500 generates a demodulation result as the square root of the sum of the squares of the IIR filter outputs.

[0130] In step S608, the processor 500 performs envelope detection to obtain the norm ||x(kT)|| of the motor signal x(kT). In a preferred variant, the motor signal is the current flowing in one phase of the motor, and the current flowing in the other phases of the motor is also detected in step S601, and the norm is obtained as the square root of the sum of the squares of the currents flowing in all phases.

[0131] In step S609, the processor 500 obtains an abnormal harmonic model from the norm ||x(kT)|| and the fundamental frequency f s (kT). The processor 500 obtains a proportional ratio R(kT) between the abnormal harmonic level and the level of abnormality in the motor. In the first embodiment of the present invention, R(kT)=A*f s (kT) B *||x(kT)|| C where A, B, and C are constants. In the second embodiment of the present invention, R(kT)=A(1 + B*f s (kT)), where A and B are constants.

[0132] In step S610, the processor 500 performs angular resampling of the proportional ratio R(kT) to obtain proportional samples R(nΔθ) using regular angular sampling.

[0133] In step S611, the processor 500 filters the samples R(nΔθ). This filtering is the notch filtering disclosed in FIG. 1. This filtering is the same as the filtering performed in step S606.

[0134] In step S612, the processor 500 performs filtering of the samples provided in the filtering step S610. This filtering is the same low-pass filtering as the filtering performed in step S606.

[0135] In step S613, the processor 500 obtains an estimated level of motor abnormality by comparing the demodulated signal obtained in step S607 with the signal derived from the signal provided by the motor and from the output of the filtering step S612, as disclosed in FIG. 1.

[0136] Thereafter, the processor 500 returns to step S601.

[0137] In one variant, the level of motor abnormality estimated in step S612 is averaged over successive signal steps.

[0138] In one variant, steps S608 to S613 are only executed when the fundamental frequency identified in step S602 exceeds a predetermined level. For example, this predetermined level is equal to the value obtained by dividing the maximum rotational speed of the motor by 2.

[0139] FIG. 7A shows a first example of spectral processing of a motor signal when an abnormality exists in the motor according to the present invention.

[0140] The horizontal axis (f / f s ) represents the signal frequency normalized with respect to the fundamental frequency, while the vertical axis (level) represents the level of the signal frequency component. The motor signal includes a strong fundamental component at a value of 1 on the horizontal axis, and a relatively small abnormal signature signal is arranged at a value of 1 - f r / f s on the horizontal axis.

[0141] This figure shows a normalized suppression band x centered on the strong fundamental component.n = f n / f s also represents the frequency response of a notch filter having. The notch filter is intended to suppress all frequency components within the suppression band. The reaction time of the notch filter is determined by the filter band. Away from the notch band, the notch filter has a single frequency response and leaves the other frequency components of the signal unchanged. Coefficient x n is selected such that the abnormal signature frequency is retained in the unprocessed region.

[0142] This figure also shows the position of the normalized center frequency x m along with the frequency band to which heterodyne non-coherent demodulation is applied according to the present invention. This frequency band includes frequencies between (x m - x c ) * f s and (x m + x c ) * f s up to.

[0143] According to the present invention, heterodyne non-coherent demodulation outputs the level of the frequency component of this abnormal signature as long as the frequency of the abnormal signature belongs to the above frequency band.

[0144] By implementing this notch filter, interference caused by strong adjacent fundamental components existing in the vicinity of the frequency band that would occur without the notch filter is suppressed, even though there is a low-pass filter having a normalized cut-off frequency x c . In this first example, the abnormal signature represents an eccentric anomaly. The coefficients (x n , x m , x c ) are selected to detect the lower sideband harmonics resulting from the eccentric anomaly. The machine has two pole pairs, positive torque is applied by the load, and the frequency of the abnormal signature signal is the frequency 1 - f m = 1 / 2 on the right side of the center frequency x s / f r .

[0145] FIG. 7B shows a second example of spectral processing of a motor signal when an abnormality exists in a motor according to the present invention.

[0146] In this second example, the abnormal signature represents an eccentricity abnormality. Coefficients (x n , x m , x c ) are selected to detect the upper sideband harmonics resulting from the eccentricity abnormality. The machine has two pole pairs, a negative torque is applied by the load, and the frequency of the abnormal signature signal is the frequency 1 + f m on the right side of the center frequency x s = 3 / 2, which is r .​

Claims

1. A method for detecting an abnormality of a motor during a transient operation of the motor, comprising: detecting at least one motor signal; identifying a fundamental frequency of the motor signal, the fundamental frequency being a frequency of a voltage waveform for driving a stator of the motor; performing heterodyne non-coherent demodulation of the motor signal in a frequency band, a center frequency and a bandwidth of the frequency band having a fixed proportional ratio with the identified fundamental frequency; determining a level of abnormality in the motor by comparing a heterodyne non-coherently demodulated signal with a signal derived from the motor signal; A method, characterized by comprising the above steps.

2. Performing the heterodyne non-coherent demodulation includes: mixing the motor signal with a sine wave and a cosine wave oscillating at the center frequency; low-pass filtering the mixed signal using a cut-off frequency equal to the bandwidth of the frequency band; combining the low-pass filtered signals; The method according to claim 1, characterized by comprising the above steps.

3. Steps performed before the heterodyne non-coherent demodulation, including: notch filtering the motor signal, a center frequency of the notch filter being equal to the identified fundamental frequency, a width of the notch having a fixed proportional ratio with the fundamental frequency, and a center frequency of the frequency band being outside a band of the notch filter; The method according to claim 2, further characterized by comprising the above step.

4. The proportional ratios of the center frequency of the frequency band, the bandwidth of the frequency band, and the center frequency of the notch filter to the fundamental frequency are determined in advance for a given type of abnormality. The method according to any one of claims 1 to 3.

5. Steps performed before the heterodyne non-coherent demodulation and / or the notch filtering, including: angle resampling the motor signal, the angle resampled signal having an equal phase increment of a signal rotating at the identified fundamental frequency; The method according to any one of claims 1 to 3, further characterized by comprising the above step.

6. The method according to claim 5, characterized in that the notch filtering and the low-pass filtering are implemented as an IIR filter having fixed predetermined coefficients.

7. Performing envelope detection on the detected motor signal; Determining a proportional ratio between the abnormal harmonic level and the level of abnormality in the motor from the envelope and the fundamental frequency; Performing angular resampling of the proportional ratio; Notch filtering the angular resampled proportional ratio; Low-pass filtering the notch filtered angular resampled proportional ratio; The method according to claim 5 or 6, further comprising the steps of:

8. A device for detecting an abnormality of a motor during a transient operation of the motor, means for detecting at least one motor signal; means for identifying a fundamental frequency of the motor signal, the fundamental frequency being a frequency of a voltage waveform for driving a stator of the motor; means for performing heterodyne non-coherent demodulation of the motor signal in a frequency band, the center frequency and the bandwidth of the frequency band having a fixed proportional ratio with the identified fundamental frequency; means for determining a level of abnormality in the motor by comparing the heterodyne non-coherent demodulated signal with a signal derived from the motor signal; A device, characterized by comprising:

Images