CN117743962A - Train traction motor bearing fault diagnosis method based on stator current signals - Google Patents
Train traction motor bearing fault diagnosis method based on stator current signals Download PDFInfo
- Publication number
- CN117743962A CN117743962A CN202311709141.6A CN202311709141A CN117743962A CN 117743962 A CN117743962 A CN 117743962A CN 202311709141 A CN202311709141 A CN 202311709141A CN 117743962 A CN117743962 A CN 117743962A
- Authority
- CN
- China
- Prior art keywords
- bearing
- motor
- imf
- frequency
- envelope
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 34
- 238000001914 filtration Methods 0.000 claims abstract description 26
- 230000003595 spectral effect Effects 0.000 claims abstract description 19
- 238000007781 pre-processing Methods 0.000 claims abstract description 13
- 238000000605 extraction Methods 0.000 claims abstract description 11
- 230000009467 reduction Effects 0.000 claims abstract description 11
- 238000012544 monitoring process Methods 0.000 claims abstract description 5
- 238000001228 spectrum Methods 0.000 claims description 53
- 238000004458 analytical method Methods 0.000 claims description 19
- 238000010606 normalization Methods 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 7
- 238000013461 design Methods 0.000 claims description 6
- 238000009825 accumulation Methods 0.000 claims description 3
- 230000036541 health Effects 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 5
- 238000004422 calculation algorithm Methods 0.000 description 13
- 238000005096 rolling process Methods 0.000 description 9
- 238000004364 calculation method Methods 0.000 description 7
- 238000009826 distribution Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 3
- 238000012952 Resampling Methods 0.000 description 2
- 230000032683 aging Effects 0.000 description 2
- 238000005260 corrosion Methods 0.000 description 2
- 230000007797 corrosion Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 238000010187 selection method Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 230000001052 transient effect Effects 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 230000003137 locomotive effect Effects 0.000 description 1
- 238000005461 lubrication Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000011946 reduction process Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000007789 sealing Methods 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Landscapes
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention relates to a rail transit high-speed motor train unit traction motor bearing fault diagnosis method, in particular to a train traction motor bearing fault diagnosis method based on stator current signals, which comprises the steps of firstly obtaining motor bearing temperature, rotating speed and motor stator current signals; when the temperature of the motor bearing reaches an alarm threshold value, bearing fault alarm is directly carried out; otherwise, monitoring the rotating speed and the current of the motor; when the rotating speed reaches a set value, sequentially carrying out data preprocessing, variation modal decomposition and optimal component reconstruction on the original data of the motor current signal, carrying out secondary filtering noise reduction and Hilbert envelope demodulation on the reconstructed signal based on the maximum spectral kurtosis, realizing weak fault feature extraction, and finally completing fault early warning and diagnosis according to threshold dynamic learning. The invention improves the traditional MCSA technology, realizes weak fault feature extraction, and effectively diagnoses early faults of the bearing.
Description
Technical Field
The invention relates to a rail transit high-speed motor train unit traction motor bearing fault diagnosis method, in particular to a train traction motor bearing fault diagnosis method based on stator current signals.
Background
On one hand, the traction motor of the rail transit high-speed motor train unit bears higher radial load for a long time and is affected by the vibration impact of complex road conditions, and the bearings are easy to form cracks at the defect positions, so that deep stripping, fracture and other fault phenomena are developed; on one hand, because the bearing is influenced by the PWM power supply mode of the three-phase alternating current inverter, the bearing is easy to generate an electric corrosion fault phenomenon due to the inherent common mode voltage generated by the bearing, namely, because the interior of the bearing is influenced by the shaft voltage, an oil film established at the contact part between a rolling bearing raceway and a rolling body can be broken down when the bearing passes through the bearing, spark discharge is generated, the contact surface is locally melted and damaged, an electric corrosion pit is formed, and finally friction is increased and abrasion is accelerated when the bearing runs; on the other hand, due to poor sealing or contaminated lubrication medium, external impurities enter between the working surfaces of the rolling bearings, so that plow-shaped scratches are formed on the working surfaces of the bearings, and the running vibration of the bearings is increased, the temperature is increased and the rotation precision is lost. Therefore, compared with the bearing with a common purpose, the rail transit traction motor bearing is more prone to failure, the common occurrence rate is about 48% of the failure of the traction motor, and the early failure diagnosis of the traction motor bearing is very important in research significance by considering the problems of stable running of a train, personal safety and the like.
The fault diagnosis technology for the motor bearing is very mature in recent years, but most of the fault diagnosis technology is mainly aimed at vibration signals, and because the vibration signals are used for diagnosing mechanical faults directly and obviously, related papers, patents and data are very rich and various, and quite a lot of documents are fault feature extraction of the bearing based on sound signals, but TCU can not acquire the signals, and an online monitoring method based on the temperature threshold of the motor bearing can reflect bearing faults, but can only appear when the service life of the bearing tends to be finished.
An authorization method for diagnosing faults of permanent magnet motor based on position sensor is disclosed in the prior art as being published in the university of Zhejiang in 2019, 01 and 08, and is disclosed in CN 106680716B. The patent is a permanent magnet motor bearing fault diagnosis method based on a position-free sensor. Firstly, voltage and current are obtained from a motor control chip, a rotor position angle and a rotating speed of a motor are obtained by using a position-free sensor algorithm, then alternating current components in the rotating speed are extracted through a moving average filter, angular domain resampling is carried out on the alternating current components according to the position angle, finally, frequency domain analysis is carried out on resampling speed signals, and whether bearing faults occur is judged according to frequency domain information. However, the patent is difficult to extract bearing fault characteristics from current signals effectively, relies on a position-free sensor, is effective only for permanent magnet motors, and cannot be diagnosed for the bearing state of asynchronous motors.
Another prior art is the application of Jiangsu university at 10 month 29 of 2021 and published at 3 month 1 of 2022, and the invention with publication number CN114112396A applies for a method for diagnosing bearing faults under the quasi-stable working condition of rotation speed fluctuation by adopting current signal analysis. The invention discloses a bearing fault diagnosis method under a rotating speed fluctuation quasi-stable working condition by adopting current signal analysis. Firstly, suppressing current measurement errors and current harmonic components caused by nonlinearity of an inverter by respectively connecting corresponding quasi-resonance compensators in parallel in a current loop and a rotating speed loop of a vector control system, secondly, designing an optimal bandpass filter by using a rapid spectrum kurtosis algorithm to extract transient impact fault components of a current signal, and then extracting a square envelope of a filtered current signal by using Hilbert transform, and further, obtaining a square envelope spectrum by using discrete Fourier transform to realize accurate extraction of fault characteristic components of a permanent magnet synchronous bearing. The invention adopts the characteristic analysis of the motor current signal, and simultaneously utilizes the characteristics of high gain and wide bandwidth of the quasi-resonant controller to realize the diagnosis of the permanent magnet synchronous motor bearing fault under the quasi-stable working condition of rotation speed fluctuation. However, the patent strongly depends on a current loop and a rotating speed loop of a motor vector control algorithm, and the used AR model filtering algorithm is complex and has more adjustable parameters; the rapid spectral kurtosis algorithm needs repeated filtering to calculate the frequency spectrum, then calculates the spectral kurtosis, is time-consuming, and has no effective filtering and noise reduction effects on a plurality of harmonic waves of current in a PWM modulation power supply mode.
The motor current characteristic analysis-based motor fault diagnosis technology is called a motor current characteristic analysis method MCSA (Motor Current Signature Analysis), and the MCSA fault analysis method is paid attention to in ten years, but when the MCSA detects the motor stator current characteristic frequency, the motor system is complex, the harmonic wave is more, the signal to noise ratio is lower in the working environment, the damage characteristic is difficult to identify, and the fault judgment value is not easy to set, so that the decision is inaccurate.
Disclosure of Invention
The invention provides a train traction motor bearing fault diagnosis method based on a stator current signal, aiming at the problem that the damage characteristic is difficult to identify when the MCSA fault analysis method detects the stator current characteristic frequency of a motor, so that the decision is inaccurate.
The invention is realized by adopting the following technical scheme: a train traction motor bearing fault diagnosis method based on stator current signals comprises the following steps: acquiring a motor bearing temperature, a motor rotating speed and a motor stator current signal; when the temperature of the motor bearing reaches an alarm threshold value, bearing fault alarm is directly carried out; otherwise, monitoring the rotating speed and the current of the motor; when the rotating speed reaches a set value, sequentially carrying out data preprocessing, variation modal decomposition and optimal component reconstruction on the original data of the motor current signal, carrying out secondary filtering noise reduction and Hilbert envelope demodulation on the reconstructed signal based on the maximum spectral kurtosis, realizing weak fault feature extraction, and finally completing fault early warning and diagnosis according to threshold dynamic learning.
According to the train traction motor bearing fault diagnosis method based on the stator current signals, the data preprocessing comprises normalization processing, band-stop filtering and direct current component removal, and the band-stop filtering is to design a band-stop filter to filter fundamental wave components of a power supply.
The train traction motor bearing fault diagnosis method based on the stator current signal comprises the following steps of: the variation modal decomposition is determined by continuously performing iterative search and optimization on the center frequency and bandwidth of each decomposition component of the motor stator current signal after data preprocessing, and decomposing the signal into K sub-signals IMF occupying different center frequency components.
The process for determining the optimal decomposition number K is as follows: firstly, determining the maximum value K_max of the decomposition number K, carrying out variation modal decomposition on a motor stator current signal from 3, and carrying out anti-modal aliasing analysis on the decomposed subcomponents; the specific anti-modal aliasing analysis method comprises the following steps: performing fast Fourier transform on each sub-component to obtain each sub-component spectrum, performing normalization processing on each sub-component spectrum to obtain a normalized spectrum, counting the current harmonic frequency values corresponding to the first M higher in amplitude in the normalized spectrum, and if M/2 of the statistical frequency values of the two adjacent sub-components are the same, adding one to K, and continuing to perform variation modal decomposition and anti-modal aliasing analysis; and if not, stopping the variation modal decomposition to determine a K value, and if K is larger than or equal to K_max, stopping the variation modal decomposition, wherein the current K value is the optimal selection value.
The train traction motor bearing fault diagnosis method based on the stator current signal comprises the following steps of: carrying out Hilbert envelope demodulation on the decomposed sub-signals IMF, obtaining an envelope IMF_env of each sub-signal IMF, carrying out zero-mean normalization processing on each envelope IMF_env, and calculating an envelope entropy E A : performing fast Fourier transform on the normalized envelope line IMF_env to obtain a sub-signal envelope spectrum IMF_fft according to the maximum characteristic frequency f of bearing faults d_max Dividing each subcomponent envelope spectrum IMF_fft into N regions for width, calculating for each according to the following formulaEnergy E of region j Then, the envelope spectrum energy entropy E of each IMF is calculated B : selecting envelope entropy E A Minimal IMF component imf_e A Envelope spectrum energy entropy E B Maximum IMF component imf_e B And E A /E B Minimal IMF component imf_e a E B And performing time domain signal reconstruction.
The train traction motor bearing fault diagnosis method based on the stator current signal comprises the following secondary filtering noise reduction process based on the maximum spectral kurtosis: performing FFT (fast Fourier transform) on the time domain signal after the optimal IMF component is reconstructed, obtaining a component frequency spectrum, movably segmenting the component frequency spectrum by using a sliding window in the frequency spectrum, obtaining a frequency band with the maximum spectral kurtosis index, using the center frequency and the bandwidth of the frequency band as indexes for designing a band-pass filter, completing the design of the band-pass filter, realizing the secondary filtering noise reduction treatment on the time domain signal after the optimal IMF component is reconstructed, filtering low-frequency components and noise frequency band components in a motor stator current signal, and reserving a medium-high frequency harmonic pulse carrier frequency band with obvious impulse of the signal.
The train traction motor bearing fault diagnosis method based on the stator current signal comprises the following steps of: performing Hilbert transform on the time domain signal after secondary filtering to obtain an envelope demodulation signal, performing FFT (fast Fourier transform) on the envelope demodulation signal to obtain an envelope spectrum, setting a frequency range by taking each bearing fault characteristic frequency in the current signal as a center, obtaining the maximum envelope spectrum amplitude in the set frequency range, and calculating an envelope spectrum amplitude average value in the set frequency range; making the continuous C times of accumulation quantity of the ratio between the two values be the diagnosis standard of the bearing state; and finally, judging and learning a dynamic threshold value by using the diagnosis standard to diagnose the health state of the motor bearing, outputting a bearing state fault if the diagnosis standard exceeds an alarm threshold value, and otherwise, outputting a bearing state to be normal.
The technical scheme of the invention has the beneficial effects that: under the condition that a vibration acquisition sensor is not additionally arranged on the motor and the converter, the active operation and maintenance PHM board card of the traction control unit TCU is used as a hardware carrier, and motor bearing fault diagnosis is carried out based on motor current, so that the problems that the traditional MCSA is easy to be subjected to environmental interference, difficult to identify damage characteristics, difficult to identify early faults, inaccurate decision and the like are solved. The invention improves the traditional MCSA technology, realizes weak fault feature extraction, and effectively diagnoses early faults of the bearing. And the diagnosis result is transmitted to the TCU, and the TCU performs on-line isolation or cutting of the motor with higher bearing fault severity, so that the running safety and reliability are ensured, and the personal safety is very significant.
Drawings
FIG. 1 is a flow chart of a bearing fault diagnosis algorithm of the present invention.
Fig. 2 is a bearing fault diagnosis envelope spectrum.
Detailed Description
Aiming at the distribution characteristics of the information acquisition system of the traction motor bearings of the motor train unit, under the condition that a vibration acquisition sensor is not additionally arranged on a motor and a converter, the active operation and maintenance PHM board card of the traction control unit TCU is used as a hardware carrier, so that the fault diagnosis of the traction motor bearings based on motor current is realized.
The invention designs a bearing fault diagnosis algorithm based on motor current: the PHM board card obtains motor bearing temperature, rotating speed and stator current signals from the traction control unit. When the temperature of the motor bearing reaches an alarm threshold value, bearing fault alarm is directly carried out; otherwise, the motor speed and current are monitored. When the rotating speed reaches a set value, sequentially carrying out data preprocessing on the original data of the motor stator current signal, carrying out variation modal decomposition and optimal component reconstruction, and carrying out secondary filtering noise reduction and Hilbert envelope demodulation on the reconstructed signal based on the maximum spectral kurtosis to realize weak fault feature extraction. And finally, completing fault early warning and diagnosis according to threshold dynamic learning, and sending a diagnosis result to the PHM unit or the TCU of the whole vehicle through a corresponding transmission protocol.
Bearing failure mechanism
The failure characteristic frequency under the vibration signal of the rolling bearing is calculated as follows:
bearing outer ring:
bearing inner race:
bearing rolling element:
bearing retainer:
wherein f r For the rotation frequency of the motor bearing, z is the number of balls, D b Is the diameter of the ball, D c Ball distribution diameter, β is contact angle.
When the bearing fails, the rotor can radially displace when passing through the failure point, so that the length of an air gap between the rotor and the stator is changed, the magnetic field distribution is changed, weak periodic pulse signals are displayed in the current of the motor stator, and the failure diagnosis of the motor bearing is performed by extracting the failure characteristics. According to the 'relation between stator current and mechanical faults' of section 2.4 of the 'failure diagnosis of locomotive traction motor bearing based on stator current analysis' of the 'Grant' graduation paper of Beijing university of transportation, the characteristic frequency of bearing faults in the current signal can be obtained through derivation of an air gap eccentric model under the bearing faults:
bearing outer ring: f (f) eccor ＝f i ±nf out
Bearing inner race: f (f) eccir ＝f i ±f r ±nf in 、f eccir ＝f i ±nf in
Bearing rolling element: f (f) eccball ＝f i ±f cage ±nf b 、f eccball ＝f i ±nf b
Wherein n=1, 2., f i For the fundamental wave frequency of the power supply。
Data preprocessing
Because of the influence of uncontrollable factors such as rich harmonic waves and the like in the traction motor power supply, data preprocessing is needed before the current signal is subjected to spectrum analysis. The preprocessing mainly comprises normalization processing, band-stop filtering and direct current component removal, and effective and reliable data support is provided for a subsequent eigenvalue extraction algorithm.
(1) Normalization
The data normalization processing adopts a min-max normalization method, and is processed according to the following formula, and the processed data is between [0,1], so that the data features have the same measurement scale, and the accuracy of the classification algorithm and the convergence speed of the optimizing algorithm are conveniently improved.
(2) Band reject filtering
When the MCSA detects the characteristic frequency of the motor stator current, because the PWM power supply is influenced by rich harmonic waves, the motor itself and other uncontrollable factors, some characteristic signals with relatively small amplitude are easily submerged in a harmonic frequency spectrum, and particularly when the damage degree is relatively small, the characteristics are less easily extracted and found. Therefore, a band-stop filter is designed to filter out the fundamental wave component of the PWM power supply.
VMD decomposition and optimal component reconstruction
The variational modal decomposition VMD is determined by continuously performing iterative search and optimization on the center frequency and bandwidth of each decomposition component of the original signal, and decomposing the signal into K sub-signals occupying different center frequency components. Specific VMD decomposition process refers to "VMD algorithm" section 1.1 of paper "gearbox State monitoring based on variational Modal decomposition"
(1) K value selection method
The number K of decomposed sub-signals during VMD decomposition is selected to have a significant effect on the decomposed sub-signals. Considering the calculation force problem of the active operation PHM board card of the traction control unit TCU and the aging problem of the diagnosis algorithm, the maximum value K_max of the decomposition layer number K is determined, otherwise, the operation time of the algorithm is too long, and the aging of the diagnosis cannot be guaranteed. In addition, considering the relation between the PWM modulation strategy and the harmonic component of the current, obtaining current data through motor operation tests under each PWM modulation strategy, performing VMD (virtual digital model) decomposition on various current data from 3, and performing anti-modal aliasing analysis on the decomposed subcomponents.
The specific anti-modal aliasing analysis method comprises the following steps: performing fast Fourier transform on each sub-component of the VMD to obtain each sub-component frequency spectrum, performing normalization processing on each sub-component frequency spectrum to obtain a normalized frequency spectrum, counting the current harmonic frequency values corresponding to the first M higher in amplitude in the normalized frequency spectrum, and if M/2 of the statistical frequency values of the two adjacent sub-components are the same, adding one to K, and continuing VMD decomposition and anti-modal aliasing analysis; otherwise stopping VMD decomposition to determine K value, if K is greater than or equal to K_max, stopping VMD, and the current K value is the optimal selection value.
(2) Envelope entropy
And carrying out VMD decomposition of a K layer on the stator current data after data preprocessing, carrying out Hilbert envelope demodulation on the decomposed sub-signals IMF by referring to Hilbert demodulation method diagnosis asynchronous motor bearing faults in section 2.1 of paper, and obtaining an envelope curve IMF_env of each sub-signal IMF.
After zero-mean normalization processing is carried out on each envelope IMF_env, the envelope entropy E is calculated according to the following formula A ：
Where L represents the length of the envelope imf_env.
(3) Envelope spectrum energy entropy
Performing fast Fourier transform on the normalized envelope to obtain a sub-signal envelope spectrum IMF_fft according to the maximum characteristic frequency f of bearing faults d_max Dividing each subcomponent envelope spectrum IMF_fft into N regions for width, calculating energy E for each region according to the following formula j Then, the envelope spectrum energy entropy E of each IMF is calculated B ：
(4) VMD optimal subcomponent selection method
The sparsity of the signal is reflected by the envelope entropy, and the more obvious the periodic pulse fault characteristic signal appears in the IMF, the larger the sparsity characteristic of the signal is, and the smaller the envelope entropy is. The energy entropy can reflect the confusion degree of the signals, bearing faults occur, fault information is reflected to the change of the spectrum energy distribution, and the higher the fault degree is, the higher the confusion degree is, and the larger the energy entropy is.
Thus, based on the above principle, the envelope entropy E is selected A Minimal IMF component imf_e A Envelope spectrum energy entropy E B Maximum IMF component imf_e B And E A /E B Minimal IMF component imf_e a E B And performing time domain signal reconstruction.
Adaptive filtering noise reduction based on maximum spectral kurtosis
The spectral kurtosis calculation formula can be used as a fourth-order accumulation amount of energy normalization, which can be understood as a peak point of a signal probability density function at each frequency, so that the spectral kurtosis can effectively reflect transient components, namely non-Gaussian components, existing in a signal, and therefore a frequency band with the maximum spectral kurtosis index is selected for fault feature extraction.
And carrying out FFT (fast Fourier transform) on the time domain signal after the optimal IMF component is reconstructed, obtaining a component frequency spectrum, and movably segmenting the component frequency spectrum into segments by using a sliding window to obtain the spectral kurtosis. And then acquiring a frequency band with the maximum spectral kurtosis index, wherein the center frequency and the bandwidth of the frequency band are used as indexes for designing a band-pass filter, the design of the band-pass filter is completed, the secondary filtering noise reduction treatment of the optimal IMF component is realized, the low-frequency component and the noise frequency band component in the motor current signal are filtered, and the medium-high frequency harmonic pulse carrier frequency band with obvious impulse pulse of the signal is reserved.
Determination of diagnostic criteria based on envelope spectra
And performing Hilbert transform on the time domain signal after the secondary filtering to obtain an envelope demodulation signal. And carrying out FFT conversion on the envelope demodulation signal to obtain an envelope spectrum. Taking account of sensor acquisition errors, bearing size errors and the like, the fault characteristic frequency f of each bearing in the current signal is used for di (f di ＝f eccor 、f eccir 、f eccball ) Setting a frequency range for the center, selecting the maximum envelope spectrum amplitude Max_f in the frequency range di The method comprises the steps of carrying out a first treatment on the surface of the And calculates an average Mean value mean_f of the envelope spectrum amplitude in the frequency range di The method comprises the steps of carrying out a first treatment on the surface of the To avoid accidental faults, the ratio between the two values Max_f is set to di /Mean_f di Is the diagnostic criteria for the bearing condition.
And finally, judging and learning a dynamic threshold value according to a diagnosis standard of the bearing state under the available fault characteristic frequency to diagnose the health state of the motor bearing, outputting the bearing state fault if the diagnosis standard exceeds an alarm threshold value, otherwise, outputting the bearing state normal.
Examples
The active operation and maintenance PHM board card is integrated in a traction control unit TCU chassis in the form of a 3U standard board card, the board card adopts a high-performance high-computation-power processor, a backboard bus interacts with the TCU through an Ethernet interface, and data such as the rotating speed of a motor, the stator current, the bearing temperature and the like are acquired in real time to carry out bearing online fault diagnosis.
Description of the examples parameters
(1) The calculation formula of the fault characteristic frequency under the vibration signal of the rolling bearing can be known:
| project | Parameters (parameters) |
| Characteristic frequency f of vibration fault of outer ring out | 6.93f r |
| Frequency f of vibration characteristic fault of inner ring in | 9.07f r |
| Characteristic frequency f of vibration fault of rolling body b | 3.08f r |
| Characteristic frequency f of vibration fault of retainer cage | 0.43f r |
(2) One of the diagnostic conditions is: the rotating speed is 2700r/min, and the torque is rated torque.
When the motor rotation speed is stabilized at 2700r/min, calculating the current fundamental frequency f of the motor current signal i 91.5Hz.
(3) Current fault characteristic frequency calculation
According to a fault characteristic frequency calculation formula under a bearing current signal, calculating a current fault characteristic frequency when n=1 as follows:
| project | Parameters (parameters) |
| Characteristic frequency f of outer ring failure eccor | 219.5、404.5 |
| Inner race characteristic failure frequency f eccir | 270.5、361.5、455.5、546.5、315.5、500.5 |
| Characteristic frequency f of rolling element failure eccball | 26.7、46.2、65.7、210.1、232.2、250.7 |
(4) The number K of the VMD decomposition subcomponents and the value of the diagnostic criteria adjustable parameter C are determined.
Description of Algorithm implementation
When the motor rotation speed is stabilized at 2700r/min, the original data of the normal motor and the fault motor are subjected to data preprocessing. Performing VMD decomposition, performing envelope entropy calculation and envelope spectrum energy entropy calculation on each subcomponent, and selecting envelope entropy E A Minimum IMF component, envelope spectral energy entropy E B And (3) reconstructing the VMD component time domain signal by using the largest IMF component, obtaining a frequency spectrum of the reconstructed time domain signal, and obtaining the spectral kurtosis in a movable sectional manner by using a sliding window. And selecting a frequency band with the maximum spectral kurtosis index to realize the secondary filtering noise reduction treatment of the reconstructed time domain signal. The Hilbert transform is carried out on the time domain signal after the secondary filtering to obtain an envelope demodulation signal, and as shown in figure 2, each fault characteristic frequency f can be selected di Is a maximum envelope spectrum amplitude within a certain range, f di At 210.1, 219.5, 232.2, 250.7, 270.5, 315.5, 361.5, 404.5, 455.5, 500.5 and 546.5Hz, respectively, the magnitude of the envelope spectrum of the fault motor is large, and the fault characteristics are obvious.
Claims (7)
1. A train traction motor bearing fault diagnosis method based on stator current signals is characterized by comprising the following steps: acquiring a motor bearing temperature, a motor rotating speed and a motor stator current signal; when the temperature of the motor bearing reaches an alarm threshold value, bearing fault alarm is directly carried out; otherwise, monitoring the rotating speed and the current of the motor; when the rotating speed reaches a set value, sequentially carrying out data preprocessing, variation modal decomposition and optimal component reconstruction on the original data of the motor current signal, carrying out secondary filtering noise reduction and Hilbert envelope demodulation on the reconstructed signal based on the maximum spectral kurtosis, realizing weak fault feature extraction, and finally completing fault early warning and diagnosis according to threshold dynamic learning.
2. The method for diagnosing the bearing faults of the traction motor of the train based on the stator current signals as claimed in claim 1, wherein the data preprocessing comprises normalization processing, band-stop filtering and direct current component removal, and the band-stop filtering is to design a band-stop filter to filter fundamental wave components of a power supply.
3. The method for diagnosing the bearing failure of the traction motor of the train based on the stator current signals as claimed in claim 2, wherein the process of the variation modal decomposition is as follows: the variation modal decomposition is determined by continuously carrying out iterative search and optimization on the center frequency and bandwidth of each decomposition component of the motor stator current signal after the data preprocessing, and decomposing the signal into components occupying different center frequenciesKSub-signals IMF.
4. A method for diagnosing a bearing failure of a traction motor of a train based on a stator current signal as claimed in claim 3, wherein the process of determining the optimal decomposition number K is: determining the number of decompositionsKMaximum value of (2)K_max， KPerforming variation modal decomposition on the motor stator current signal from 3, and performing anti-modal aliasing analysis on the decomposed subcomponents; the specific anti-modal aliasing analysis method comprises the following steps: performing fast Fourier transform on each sub-component to obtainTo each sub-component spectrum, carrying out normalization processing on each sub-component spectrum to obtain a normalized spectrum, and counting the front part with higher amplitude in the normalized spectrumMThe corresponding current harmonic frequency value is that if the statistical frequency values of the adjacent two subcomponents are included inM/2If the number is the same, adding one to K, and continuing to perform variation modal decomposition and anti-modal aliasing analysis; otherwise stopping the variational modal decomposition determinationKValue of ifKGreater than or equal toK_maxThe decomposition of the variation mode is stopped at the momentKThe value is the best choice value.
5. The method for diagnosing a bearing failure of a traction motor of a train based on a stator current signal as recited in claim 4, wherein the process of reconstructing the optimal component is as follows: carrying out Hilbert envelope demodulation on the decomposed sub-signals IMF to obtain an envelope curve of each sub-signal IMFIMF_envFor each envelope curveIMF_envAfter zero mean normalization processing, calculating the envelope entropyE A : for the envelope curve after normalizationIMF_envPerforming fast Fourier transform to obtain sub-signal envelope spectrumIMF_fftAccording to the maximum characteristic frequency of bearing faultsf d_max Envelope spectrum of each subcomponent for widthIMF_fftDividing into N regions, and calculating energy for each region according to the following formulaE j Then, the energy entropy of the envelope spectrum of each IMF is calculatedE B : selecting envelope entropyE A Minimal IMF componentIMF_E A Envelope spectrum energy entropyE B Maximum IMF componentIMF_E B And (3) the methodE A /E B Minimal IMF componentIMF_E a E B And performing time domain signal reconstruction.
6. The method for diagnosing the bearing fault of the traction motor of the train based on the stator current signal as recited in claim 5, wherein the process of secondary filtering and noise reduction based on the maximum spectral kurtosis is as follows: performing FFT (fast Fourier transform) on the time domain signal after the optimal IMF component is reconstructed, obtaining a component frequency spectrum, movably segmenting the component frequency spectrum by using a sliding window in the frequency spectrum, obtaining a frequency band with the maximum spectral kurtosis index, using the center frequency and the bandwidth of the frequency band as indexes for designing a band-pass filter, completing the design of the band-pass filter, realizing the secondary filtering noise reduction treatment on the time domain signal after the optimal IMF component is reconstructed, filtering low-frequency components and noise frequency band components in a motor stator current signal, and reserving a medium-high frequency harmonic pulse carrier frequency band with obvious impulse of the signal.
7. The method for diagnosing faults of a train traction motor bearing based on stator current signals as claimed in claim 6, wherein the process of achieving weak fault feature extraction is realized by Hilbert envelope demodulation: performing Hilbert transform on the time domain signal after secondary filtering to obtain an envelope demodulation signal, performing FFT (fast Fourier transform) on the envelope demodulation signal to obtain an envelope spectrum, setting a frequency range by taking each bearing fault characteristic frequency in the current signal as a center, obtaining the maximum envelope spectrum amplitude in the set frequency range, and calculating an envelope spectrum amplitude average value in the set frequency range; let the ratio between the two values be continuousCThe secondary accumulation is a diagnostic criterion for the state of the bearing; and finally, judging and learning a dynamic threshold value by using the diagnosis standard to diagnose the health state of the motor bearing, outputting a bearing state fault if the diagnosis standard exceeds an alarm threshold value, and otherwise, outputting a bearing state to be normal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311709141.6A CN117743962A (en) | 2023-12-13 | 2023-12-13 | Train traction motor bearing fault diagnosis method based on stator current signals |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311709141.6A CN117743962A (en) | 2023-12-13 | 2023-12-13 | Train traction motor bearing fault diagnosis method based on stator current signals |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117743962A true CN117743962A (en) | 2024-03-22 |
Family
ID=90253733
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311709141.6A Pending CN117743962A (en) | 2023-12-13 | 2023-12-13 | Train traction motor bearing fault diagnosis method based on stator current signals |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117743962A (en) |
-
2023
- 2023-12-13 CN CN202311709141.6A patent/CN117743962A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108195587B (en) | Motor rolling bearing fault diagnosis method and system | |
| Cheng et al. | Adaptive multipoint optimal minimum entropy deconvolution adjusted and application to fault diagnosis of rolling element bearings | |
| CN111089726B (en) | Rolling bearing fault diagnosis method based on optimal dimension singular spectrum decomposition | |
| Refaat et al. | ANN-based for detection, diagnosis the bearing fault for three phase induction motors using current signal | |
| CN112378633A (en) | Mechanical fault diagnosis method | |
| Jain et al. | A review on vibration signal analysis techniques used for detection of rolling element bearing defects | |
| Zhu et al. | Adaptive combined HOEO based fault feature extraction method for rolling element bearing under variable speed condition | |
| Saidi et al. | Wind turbine high-speed shaft bearing degradation analysis for run-to-failure testing using spectral kurtosis | |
| CN117743962A (en) | Train traction motor bearing fault diagnosis method based on stator current signals | |
| Gryllias et al. | A peak energy criterion (pe) for the selection of resonance bands in complex shifted morlet wavelet (csmw) based demodulation of defective rolling element bearings vibration response | |
| Zhang et al. | Fault diagnosis of axle box bearing with acoustic signal based on chirplet transform and support vector machine | |
| CN117093938A (en) | Fan bearing fault detection method and system based on deep learning | |
| Djemili et al. | A Wind Turbine Bearing Fault Detection Method Based on Improved CEEMDAN and AR-MEDA | |
| Xu et al. | Diagnosis of weak fault of rolling bearing based on EEMD and envelope spectrum analysis | |
| CN109975025B (en) | Locomotive wheel pair bearing quantitative diagnosis method based on adaptive filtering demodulation | |
| Ma et al. | Envelope demodulation method based on SET for fault diagnosis of rolling bearings under variable speed | |
| Guo et al. | Fault identification of low-speed hub bearing of crane based on MBMD and BP neural network | |
| Zhou et al. | Reinforced Morlet wavelet transform for bearing fault diagnosis | |
| CN112161805A (en) | Bearing tiny fault diagnosis method based on time series scale analysis and CNN | |
| Sonmez et al. | 851. Entropy-based fault detection approach for motor vibration signals under accelerated aging process | |
| Xia et al. | Endpoint effect suppression based on multipoint extension in bearing fault diagnosis | |
| Qijun et al. | Early fault diagnosis of the rolling bearing based on the weak signal detection technology | |
| Yang et al. | Motor Current Time-varying Quadratic Phase Coupling Analysis and its Application in Traction Motor Fault Detection under Varying-speed Condition | |
| CN111366360B (en) | Method for detecting early fault of planetary gear box by using rotating speed signal | |
| CN116985865B (en) | Method, device and system for diagnosing and detecting polygonal faults of wheels of rail transit |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |