CN111094927A - Bearing fault diagnosis method and device, readable storage medium and electronic equipment - Google Patents
Bearing fault diagnosis method and device, readable storage medium and electronic equipment Download PDFInfo
- Publication number
- CN111094927A CN111094927A CN201780094590.0A CN201780094590A CN111094927A CN 111094927 A CN111094927 A CN 111094927A CN 201780094590 A CN201780094590 A CN 201780094590A CN 111094927 A CN111094927 A CN 111094927A
- Authority
- CN
- China
- Prior art keywords
- bearing
- matrix
- fault diagnosis
- wavelet
- vibration acceleration
- 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 81
- 238000000034 method Methods 0.000 title claims abstract description 44
- 239000011159 matrix material Substances 0.000 claims abstract description 171
- 230000001133 acceleration Effects 0.000 claims abstract description 99
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 67
- 239000013598 vector Substances 0.000 claims abstract description 67
- 238000003062 neural network model Methods 0.000 claims abstract description 66
- 230000006870 function Effects 0.000 claims description 30
- 230000009467 reduction Effects 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 10
- 230000007257 malfunction Effects 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Hardware Design (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Geometry (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Acoustics & Sound (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
A bearing fault diagnosis method and device, a readable storage medium and an electronic device are provided, wherein the fault diagnosis method comprises the following steps: collecting a vibration acceleration signal of a bearing in a working state (S101); performing wavelet decomposition on a vibration acceleration signal of the bearing (S102); sequentially extracting characteristic parameters from signals obtained by wavelet decomposition and constructing characteristic vectors (S103); constructing a feature matrix according to the feature vector (S104); inputting the feature matrix to a pre-trained neural network model (S105); and comparing the output result of the neural network model with a preset target matrix to obtain a fault diagnosis result of the bearing (S106). The bearing fault diagnosis accuracy can be improved by the aid of the scheme.
Description
Technical Field
The invention relates to the field of bearing fault detection, in particular to a bearing fault diagnosis method and device, a readable storage medium and electronic equipment.
Background
The bearing is a basic part in mechanical equipment and has been widely applied to equipment such as trains and the like. The running state of the bearing plays an important role in the running safety of the train, and therefore, the diagnosis of the bearing fault is particularly important.
In the prior art, when a bearing fault is diagnosed, a data analyst usually observes an envelope spectrum waveform of a demodulated time-domain vibration acceleration signal to perform subjective judgment, and needs the data analyst with a relevant theoretical background to analyze the data, so that the subjectivity is high. Although the prior art also discloses a method for diagnosing the bearing fault by using the neural network theory, the bearing fault is diagnosed with low accuracy.
Disclosure of Invention
The technical problem solved by the embodiment of the invention is how to improve the accuracy of bearing fault diagnosis.
In order to solve the above technical problem, an embodiment of the present invention provides a bearing fault diagnosis method, including: collecting a vibration acceleration signal of a bearing in a working state; carrying out wavelet decomposition on the vibration acceleration signal of the bearing; sequentially extracting characteristic parameters from signals obtained by wavelet decomposition and constructing characteristic vectors; constructing a feature matrix according to the feature vector; inputting the feature matrix into a pre-trained neural network model; and comparing the output result of the neural network model with a preset target matrix to obtain the fault diagnosis result of the bearing.
Optionally, the fault diagnosis result of the bearing includes at least one of: whether the bearing is malfunctioning, the bearing element that is malfunctioning, and the severity of the malfunction.
Optionally, the acquiring a vibration acceleration signal of the bearing includes: and acquiring a time domain vibration acceleration signal of the bearing.
Optionally, the performing wavelet decomposition on the vibration acceleration signal of the bearing includes: selecting a wavelet mother function, wherein the waveform of the wavelet mother function is related to the waveform of the time domain vibration acceleration signal of the bearing; performing wavelet decomposition on the wavelet mother function to obtain time domain signals of N frequency bands, wherein N is 2jAnd j is the level of wavelet decomposition.
Optionally, the sequentially extracting feature parameters from the signals obtained by wavelet decomposition and constructing feature vectors includes: and sequentially extracting characteristic parameters from the frequency domain signals of the N frequency bands, and constructing a characteristic vector.
Optionally, the constructing a feature matrix according to the feature vector includes: constructing a feature matrix according to the feature vector; the characteristic matrix is an M multiplied by N matrix, and M is the number of the characteristic vectors.
Optionally, the target matrix is an M × K matrix, and K is the number of components of the bearing to be diagnosed.
Optionally, before performing wavelet decomposition on the vibration acceleration signal, the method further includes: and carrying out noise reduction processing on the vibration acceleration signal.
Optionally, the performing noise reduction processing on the vibration acceleration signal includes: and denoising the vibration acceleration signal by adopting a wavelet threshold denoising method.
Optionally, the characteristic parameter includes at least one of: the energy of each wavelet decomposed signal, the crest factor of each wavelet decomposed signal, the standard deviation of each wavelet decomposed signal, the root mean square of each wavelet decomposed signal and the 90-quantile value of each wavelet decomposed signal.
Optionally, before inputting the feature matrix into the pre-generated neural network model, the method further includes: and carrying out normalization processing on elements in the feature matrix.
Optionally, the neural network model includes: an input layer, a hidden layer, and an output layer; and the input layer data of the neural network model is the characteristic matrix, and the output layer data of the neural network model is the target matrix.
Optionally, the comparing the output result of the neural network model with a preset target matrix to obtain the fault diagnosis result of the bearing includes: comparing the output result of the neural network model with the target matrix, obtaining an error value between the output result and the target matrix, and judging that the bearing component corresponding to the minimum error value of the target matrix has a fault.
The embodiment of the invention also provides a bearing fault diagnosis device, which comprises: the acquisition unit is used for acquiring a vibration acceleration signal of the bearing in a working state; the wavelet decomposition unit is used for performing wavelet decomposition on the vibration acceleration signal of the bearing; the characteristic vector construction unit is used for sequentially extracting characteristic parameters from signals obtained by wavelet decomposition and constructing characteristic vectors; the characteristic matrix construction unit is used for constructing a characteristic matrix according to the characteristic vector; the input unit is used for inputting the characteristic matrix to a pre-trained neural network model; and the fault diagnosis result acquisition unit is used for comparing the output result of the neural network model with a preset target matrix to acquire the fault diagnosis result of the bearing.
Optionally, the fault diagnosis result of the bearing includes at least one of: whether the bearing is malfunctioning, the bearing element that is malfunctioning, and the severity of the malfunction.
Optionally, the acquisition unit is configured to acquire a time-domain vibration acceleration signal of the bearing.
Optionally, theThe wavelet decomposition unit is used for selecting a wavelet mother function, and the waveform of the wavelet mother function is related to the waveform of the time domain vibration acceleration signal of the bearing; performing wavelet decomposition on the wavelet mother function to obtain time domain signals of N frequency bands, wherein N is 2jAnd j is the level of wavelet decomposition.
Optionally, the feature vector constructing unit is configured to demodulate the time domain signals of the N frequency bands and perform time-frequency conversion to obtain frequency domain signals of the corresponding N frequency bands; and sequentially extracting characteristic parameters from the frequency domain signals of the N frequency bands, and constructing a characteristic vector.
Optionally, the feature matrix constructing unit is configured to construct a feature matrix according to the feature vector; the characteristic matrix is an M multiplied by N matrix, and M is the number of the characteristic vectors.
Optionally, the target matrix is an M × K matrix, and K is the number of components of the bearing to be diagnosed.
Optionally, the bearing fault diagnosis device further includes: and the denoising processing unit is used for denoising the vibration acceleration signal before the wavelet decomposition unit performs wavelet decomposition on the vibration acceleration signal.
Optionally, the denoising processing unit is configured to perform denoising processing on the vibration acceleration signal by using a wavelet threshold denoising method.
Optionally, the characteristic parameter includes at least one of: the energy of each wavelet decomposed signal, the crest factor of each wavelet decomposed signal, the standard deviation of each wavelet decomposed signal, the root mean square of each wavelet decomposed signal and the 90-quantile value of each wavelet decomposed signal.
Optionally, the bearing fault diagnosis device includes: and the normalization processing unit is used for normalizing the elements in the characteristic matrix before the input unit inputs the characteristic matrix to a pre-generated neural network model.
Optionally, the neural network model includes: an input layer, a hidden layer, and an output layer; and the input layer data of the neural network model is the characteristic matrix, and the output layer data of the neural network model is the target matrix.
Optionally, the fault diagnosis result obtaining unit is configured to compare an output result of the neural network model with the target matrix, obtain an error value between the output result and the target matrix, and determine that the bearing component corresponding to the minimum error value of the target matrix has a fault.
The embodiment of the present invention further provides a computer readable storage medium, on which computer instructions are stored, and when the computer instructions are executed, the steps of any one of the bearing fault diagnosis methods described above are executed.
The embodiment of the present invention further provides an electronic device, which includes a memory and a processor, where the memory stores computer instructions executable on the processor, and the processor executes the computer instructions to perform any of the steps of the bearing fault diagnosis method described above.
Compared with the prior art, the technical scheme of the embodiment of the invention has the following beneficial effects:
the method comprises the steps of performing wavelet decomposition on an acquired vibration acceleration signal of the bearing, sequentially extracting characteristic parameters from the signal obtained by the wavelet decomposition and constructing a characteristic vector, constructing a characteristic matrix according to the characteristic vector and taking the characteristic matrix as the input of a neural network model, processing the characteristic matrix through the pre-trained neural network model to obtain an output result, comparing the output result with a preset target matrix to obtain a fault diagnosis result of the bearing, wherein the judgment of the whole bearing fault does not depend on the subjective judgment of a tester, and therefore the accuracy of bearing fault diagnosis can be improved.
Furthermore, before the wavelet decomposition is carried out on the vibration acceleration signal of the bearing, the noise reduction processing is carried out on the vibration acceleration signal, and the interference of other signals on the vibration acceleration signal is reduced, so that the accuracy of bearing fault diagnosis can be further improved.
Drawings
FIG. 1 is a flow chart of a bearing fault diagnosis method in an embodiment of the invention;
FIG. 2 is a schematic diagram of a wavelet decomposition in an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a bearing fault diagnosis device in an embodiment of the present invention.
Detailed Description
In the prior art, when a bearing fault is diagnosed, a data analyst usually observes an envelope spectrum waveform of a demodulated time-domain vibration acceleration signal to carry out subjective judgment, and needs the data analyst with a relevant theoretical background to analyze the data, so that the subjectivity is high. Although the prior art also discloses a method for diagnosing the bearing fault by using the neural network theory, the bearing fault is diagnosed with low accuracy.
In the embodiment of the invention, the collected vibration acceleration signals of the bearing are subjected to wavelet decomposition, the characteristic parameters are sequentially extracted from the signals obtained by the wavelet decomposition, the characteristic vectors are constructed, the characteristic matrix is constructed according to the characteristic vectors and is used as the input of the neural network model, the characteristic matrix is processed by the pre-trained neural network model to obtain the output result, the output result is compared with the preset target matrix to obtain the fault diagnosis result of the bearing, the whole bearing fault judgment does not depend on the subjective judgment of a tester, and therefore, the bearing fault diagnosis accuracy can be improved.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below.
The embodiment of the invention provides a bearing fault diagnosis method, which is described in detail by referring to fig. 1 through specific steps.
And step S101, acquiring a vibration acceleration signal of the bearing in a working state.
In a specific implementation, when the bearing is in a working state, a vibration acceleration signal of the bearing can be extracted from a vibration signal corresponding to a mechanical system in which the bearing is located.
In practice, a bearing is usually a component of a mechanical system. When the mechanical system is in a working state, a preset acceleration sensor can be adopted to acquire the vibration signal of the mechanical system in real time. The acceleration sensor can be built in the mechanical system or can be arranged independently of the mechanical system.
The vibration signals of the mechanical system collected by the acceleration sensor include vibration acceleration signals of the bearing, vibration acceleration signals of other components in the mechanical system, and some background noise. Therefore, in the specific implementation, after the vibration signals of the mechanical system are collected by the acceleration sensor, the vibration acceleration signals of the bearing can be extracted from the vibration signals.
In a specific implementation, the vibration acceleration signal of the bearing can also be directly acquired through a sensor coupled with the bearing. For example, a sensor coupled to the bearing is preset, and when the bearing operates, a vibration acceleration signal of the bearing is acquired in real time through the sensor.
In a specific implementation, no matter the vibration acceleration signal of the bearing is extracted from the vibration signal of the mechanical system, or the vibration acceleration signal of the bearing is directly collected through a sensor coupled to the bearing, other noises may be mixed in the obtained vibration acceleration signal. Understandably, the vibration acceleration signal of the bearing collected in the application is a signal corresponding to the collected bearing under the working condition.
Therefore, after the vibration acceleration signal of the bearing in the working state is acquired, the vibration acceleration signal of the bearing can be subjected to noise reduction processing to reduce the noise in the vibration acceleration signal of the bearing. In a specific implementation, a wavelet threshold denoising method can be adopted to perform denoising processing on the vibration acceleration signal of the bearing. It can be understood that other noise reduction methods may also be adopted to perform noise reduction processing on the vibration acceleration signal of the bearing, which is not described herein again.
And step S102, performing wavelet decomposition on the vibration acceleration signal of the bearing.
In specific implementation, the collected vibration acceleration signal of the bearing may be a time-domain vibration acceleration signal of the bearing, and may also be a frequency-domain vibration acceleration signal of the bearing. And when the acquired vibration acceleration signal of the bearing is a time domain vibration acceleration signal of the bearing, performing wavelet decomposition on the time domain vibration acceleration of the bearing. When the acquired vibration acceleration signal of the bearing is a frequency domain vibration acceleration signal of the bearing, performing wavelet decomposition on the frequency domain vibration acceleration signal of the bearing.
When wavelet decomposition is carried out, if the frequency domain vibration acceleration signal of the bearing is decomposed, a wavelet mother function of which the waveform is related to the waveform of the frequency domain vibration acceleration signal of the bearing can be selected, and the wavelet decomposition is carried out on the selected wavelet mother function; if the time domain vibration acceleration signal of the bearing is decomposed, a wavelet mother function of which the waveform is related to the waveform of the time domain vibration acceleration signal of the bearing can be selected, and the selected wavelet mother function is decomposed.
The wavelet decomposition of the time-domain vibration acceleration signal of the bearing is described as an example.
In a specific implementation, a wavelet mother function of which the waveform is related to the waveform of the collected time-domain vibration signal of the bearing is selected. Performing wavelet decomposition on the selected wavelet mother function to obtain time domain signals of N frequency bands, wherein N is 2jAnd j is the level of wavelet decomposition.
In a specific application, the order of wavelet decomposition can be set according to the actual application requirement. In an embodiment of the present invention, j is set to 3, and in this case, 8 time domain signals with the same frequency bandwidth are obtained after wavelet decomposition is performed on the wavelet mother function.
Referring to fig. 2, a schematic diagram of a wavelet decomposition in an embodiment of the present invention is given. In fig. 2, the level j of the wavelet decomposition is 3.
In fig. 2, the signal S is decomposed by one layer to obtain the signal SA1Sum signal SD1. For signal SA1Decomposing to obtain signal SAA2Sum signal SDA2(ii) a For signal SD1Decomposing to obtain signal SAD2Sum signal SDD2. To pairSignal SAA2Decomposing to obtain signal S30And S31(ii) a For signal SDA2Decomposing to obtain signal S32And S33(ii) a For signal SAD2Decomposing to obtain signal S34And S35(ii) a For signal SDD2Decomposing to obtain signal S36And S37。
That is to say, after the signal S is decomposed and reconstructed by three layers of wavelets, the obtained third layer signal includes 8 time domain signals with the same frequency bandwidth, which are: s30、S31、S32、S33、S34、S35、S36And S37。
In a specific implementation, when the wavelet mother function is selected, a shape of a waveform related to a waveform of the vibration acceleration signal of the bearing may be selected, and the waveform correlation may be: the waveform of the wavelet mother function is relatively similar to the waveform of the vibration acceleration signal of the bearing, and the approximation degree of the wavelet mother function and the waveform of the vibration acceleration signal of the bearing reaches a certain threshold value.
For example, a wavelet mother function having a waveform similar to that of the time-domain vibration acceleration signal of the bearing is selected based on empirical values. Because the selected wavelet mother function is related to the waveform of the time domain vibration acceleration signal of the bearing, after the wavelet decomposition is carried out on the wavelet mother function, the signal related to the time domain vibration acceleration signal of the bearing in the time domain signals with the same N frequency bands is highlighted, and the signal unrelated to the time domain vibration acceleration signal of the bearing is weakened.
In a specific implementation, time domain signals of N frequency bands are demodulated, if the time domain vibration acceleration signal of the bearing contains impact frequency components caused by bearing faults, the impact frequency components caused by the bearing faults can be extracted from the high-frequency carrier signal after the N time domain signals are demodulated, and the characteristics of the demodulated time domain signals of the frequency bands are more obvious.
In practical applications, it can be known that after time-frequency conversion is performed on a time domain signal to obtain a frequency domain signal, the characteristics of a waveform in a frequency domain are more obvious than those of a waveform in a time domain. Therefore, after the time domain signals of the N frequency bands are demodulated, the time domain signals of the N frequency bands after the demodulation are time-frequency converted, and the time domain signals of the N frequency bands are converted into corresponding frequency domain signals of the N frequency bands.
When time-frequency conversion is performed on the time domain signals of the N frequency bands, a Fast Fourier Transform (FFT) method may be adopted to obtain frequency domain signals corresponding to the N frequency bands.
And step S103, sequentially extracting characteristic parameters from the signals obtained by wavelet decomposition and constructing characteristic vectors.
In a specific implementation, the characteristic parameters extracted from the signals obtained by wavelet decomposition may include at least one of: the energy of each wavelet decomposed signal, the crest factor of each wavelet decomposed signal, the root mean square of each wavelet decomposed signal and the 90-quantile value of each wavelet decomposed signal.
In specific implementation, after wavelet decomposition is performed on the wavelet mother function to obtain frequency domain signals of N frequency bands, the same feature parameters can be extracted from the frequency domain signals of the N frequency bands, respectively, to construct feature vectors. The number of the constructed feature vectors is the same as the number of feature parameters extracted from the frequency domain signal of each frequency band, and the number of elements in each feature vector is equal to N.
For example, 5 feature parameters are extracted from the frequency domain signal of each frequency band, and N is 8, the number of constructed feature vectors is 5, and the number of elements in each feature vector is 8.
In a specific implementation, the characteristic parameters sequentially extracted from the frequency domain signals of the N frequency bands may include at least one of the following: the energy of the frequency-domain signal of each frequency band, the crest factor of the frequency-domain signal of each frequency band, the standard deviation of the frequency-domain signal of each frequency band, the root mean square of the frequency-domain signal of each frequency band, and the 90-quantile value of the frequency-domain signal of each frequency band.
And step S104, constructing a feature matrix according to the feature vector.
In a specific implementation, after the feature vectors obtained in step S103, a corresponding feature matrix may be constructed.
In specific implementation, after wavelet decomposition is performed on the wavelet mother function to obtain frequency domain signals of N frequency bands, the same feature parameters can be extracted from the frequency domain signals of the N frequency bands, respectively, to construct feature vectors. At this time, the constructed feature matrix is an M × N matrix, and M is the number of the obtained feature vectors.
For example, when the time domain vibration acceleration signal is subjected to wavelet decomposition of j-3 layers, N-8; for the frequency domain signal of each frequency band, 5 feature parameters are extracted, and then 5 feature vectors are constructed. The feature matrix constructed is a 5 x 8 matrix, i.e. the feature matrix is a matrix of 5 rows and 8 columns.
Next, step S103 to step S104 will be described with N being 8 as an example.
In practical applications, it is known that the signal S can be reconstructed from the decomposed signal without errors according to the wavelet packet reconstruction theory. According to the principle of conservation of energy, the energy in the time domain is converted into the energy in the frequency domain by the same amount, and therefore, the following equation can be obtained:
FS＝FS30+FS31+FS32+FS33+FS34+FS35+FS36+FS37； (1)
in the formula (1), FS is a frequency domain vibration acceleration signal corresponding to the time domain vibration acceleration signal, FS30～FS37In turn a time domain signal S30～S37Corresponding frequency domain signals.
Setting the energy corresponding to the frequency domain signal of the ith frequency band as E3iThen, then
Wherein i is 0,1,2, …,7, k is 1,2, …, n, xikThird layer signal S for wavelet decomposition3iThe wavelet envelope coefficient value of (a). Thus, the total energy corresponding to the frequency domain signals of the N frequency bands is:
a certain one of the third layerThe energy of the relative wavelet packet coefficient of the frequency domain signal of each frequency band is
After obtaining the energy corresponding to the frequency domain signal of each frequency band, an energy-based feature vector can be constructed as follows: r ═ R30，R31，…，R37]
In practical applications, it is known that the peak factor of the frequency domain signal reflects the level of the vibration impact component in the frequency domain signal.
In a specific implementation, the crest factor of the frequency domain signal of the ith frequency band is:
wherein, peak3iPeak, RMS, of frequency domain signal for the ith frequency band3iIs the rms value of the frequency domain signal of the ith frequency band.
After obtaining the peak factor corresponding to the frequency domain signal of each frequency band, a feature vector based on the peak factor can be constructed as follows: c ═ C30，C31，...，C37]
In practical applications, the standard deviation reflects the degree of dispersion of data in an array. When the bearing is in fault, impact can be generated, and fault signals of various frequency bands have obvious characteristics after demodulation, so that the standard deviation can reflect the characteristics of the fault signals.
In a specific implementation, the standard deviation of the frequency domain signal of the ith frequency band is:
wherein L is the number of values of the frequency domain signal of the ith frequency band, xlIs the l-th value in the frequency domain signal of the i-th frequency band, r is the average value of the signal of the i-th frequency band
After obtaining the standard deviation of the frequency domain signal for each frequency band, it can be constructedEstablishing a characteristic vector based on the standard deviation as follows: σ ═ σ [ σ ]30，σ31，...，σ37]。
In practical applications, the root mean square can reflect the overall level of energy in the signal, and thus can be used as a characteristic parameter for bearing fault detection.
In a specific implementation, the root mean square of the frequency domain signal of the ith frequency band is:
wherein L is the number of values of the frequency domain signal of the ith frequency band, xlIs the ith value in the frequency domain signal of the ith frequency band.
After the root mean square of the frequency domain signal of each frequency band is obtained, a feature vector based on the root mean square can be constructed as follows: RMS ═ RMS30，RMS31，…，RMS37]。
In specific implementation, for the frequency domain signal of the ith frequency band, a 90-quantile value P of the maximum value of the ith frequency band is taken3i. After obtaining the 90-quantile value corresponding to the frequency domain signal of each frequency band, a feature vector based on the 90-quantile value can be constructed as follows: p ═ P30，P31，…，P37]。
It can be understood that the feature parameters extracted from the frequency domain signals of the N frequency bands are not limited to the ones provided in the above embodiments of the present invention, and other feature parameters capable of representing the features of the frequency domain signals may also exist, which are not described herein again.
In a specific implementation, one or more of the feature vectors generated in step S103 may be selected to generate a constructed feature matrix according to an actual application requirement. For example, when N is 8, the feature matrix is constructed by selecting the energy-based feature vector and the peak factor-based feature vector from the feature vectors generated in step S103, and in this case, the feature matrix is a 2 × 8 matrix. If N is 8 and all the eigenvectors generated in step S103 are selected to construct the feature matrix, the constructed feature matrix is a 5 × 8 matrix, for example.
And step S105, inputting the feature matrix into a pre-trained neural network model.
In particular implementations, neural network models can be generated in advance and trained. In practical application, a plurality of groups of data of known faults to be detected or diagnosed can be collected in advance, and the generated neural network model is trained. In the training process, the generated neural network model can continuously adjust the weight of each layer according to the input known training matrix until the error of the neural network model meets the set minimum error.
In specific implementation, in order to accelerate the convergence characteristic of the neural network model, before the feature matrix is input into the neural network model, normalization processing is performed on the feature matrix. When the feature matrix is normalized, the element with the largest value may be selected from the feature matrix, and all other elements are divided by the selected largest element, thereby implementing the normalization of the feature matrix.
And S106, comparing the output result of the neural network model with a preset target matrix to obtain a fault diagnosis result of the bearing.
Next, step S105 to step S106 will be described.
In a specific implementation, the target matrix may be pre-constructed. The pre-constructed target matrix is an M K matrix, and K is the part number of the bearing to be diagnosed.
In practice, it is known that a bearing can be generally composed of four parts, namely an Outer Ring (OR), an Inner Ring (IR), rollers (Roller), and a Cage (Cage). In practical applications, the training process of the neural network model may include the following steps: 1) preparing data; taking the extracted vibration acceleration signal, a fault signal characteristic matrix corresponding to an outer ring of the bearing, a fault signal characteristic matrix corresponding to an inner ring of the bearing, a fault signal characteristic matrix corresponding to a roller of the bearing and a fault signal characteristic matrix corresponding to a bearing retainer as input of a neural network, and training the neural network according to a set target matrix as output; 2) training a neural network model; and in the process of training the neural network model, continuously correcting the weight and the threshold of the neural network model according to the set minimum error value until the minimum error value is met.
In particular implementations, the data required for training the neural network model and the setting of the minimum error value may be guided by the fault diagnosis goals, such as whether a fault is generated, the subject of the fault diagnosis (referring to the bearing element to be diagnosed), the severity of the fault under different operating conditions, and the like.
In a specific implementation, K is set to 4. When K is 4, the target matrix may be set as follows:
in the target matrix, a first row is a target vector corresponding to an outer ring of the bearing when a fault occurs, a second row is a target vector corresponding to an inner ring of the bearing when a fault occurs, a third row is a target vector corresponding to an outer ring of a roller of the bearing when a fault occurs, and a fourth row is a target vector corresponding to a retainer of the bearing when a fault occurs.
Because the feature matrix constructed in the embodiment of the invention is used as the input of the neural network model, and the feature matrix is an M multiplied by N matrix, the target matrix can be reconstructed, and each column in the target matrix is expanded into an independent M multiplied by K matrix, so that K target matrices are obtained, wherein the K target matrices are respectively in one-to-one correspondence with the K components. For each M × K matrix, the elements of different rows therein are the same.
In an embodiment of the present invention, when K is 4 and M is 5, the 4 columns in the target matrix are respectively expanded into 4 independent 5 × 4 target matrices, which are respectively a target matrix corresponding to an outer ring of the bearing, a target matrix corresponding to an inner ring of the bearing, a target matrix corresponding to a roller of the bearing, and a target matrix corresponding to a cage of the bearing, wherein:
the target matrix corresponding to the normal state of the bearing is as follows:
target moment corresponding to outer ring of bearingThe matrix is as follows:
the target matrix corresponding to the inner ring of the bearing is as follows:
the target matrix corresponding to the rollers of the bearing is:
the target matrix corresponding to the retainer of the bearing is as follows:
in an embodiment of the present invention, the target matrix corresponding to the outer ring of the bearing, the target matrix corresponding to the inner ring of the bearing, the target matrix corresponding to the roller of the bearing, and the target matrix corresponding to the retainer of the bearing are generated by expanding a preset target matrix.
In a specific implementation, the neural network model includes an input layer, a hidden layer, and an output layer. The input layer data of the neural network model is a characteristic matrix, and the output layer data is a preset target matrix. In order to obtain the diagnosis result, the output layer data can be a target matrix corresponding to an outer ring of the bearing, a target matrix corresponding to an inner ring of the bearing, a target matrix corresponding to a roller of the bearing and a target matrix corresponding to a retainer of the bearing, wherein the target matrices are generated by target matrix expansion.
In specific implementation, the output result of the neural network model is compared with the target matrix, an error value between the output result and the target matrix is obtained, and the bearing component corresponding to the minimum error value of the target matrix is judged to have a fault.
When the output result of the neural network model is compared with the target matrix, the output result of the neural network model can be compared with the expanded target matrix respectively, namely the target matrix corresponding to the outer ring of the bearing, the target matrix corresponding to the inner ring of the bearing, the target matrix corresponding to the rollers of the bearing and the target matrix corresponding to the retainer of the bearing are compared one by one, and the minimum error value of the expanded target matrix is obtained, so that which component of the bearing fails is obtained.
For example, if the output result of the neural network model is a 5 × 4 matrix and the values in the 1 st column are all between 0.9 and 1, it can be known that the error value between the output result of the neural network model and the target matrix corresponding to the outer ring of the bearing is the minimum, and thus it is determined that the outer ring of the bearing has a fault.
Therefore, the wavelet decomposition is carried out on the collected bearing time domain vibration acceleration signals, and the high-frequency part and the low-frequency part of the time domain vibration acceleration signals are considered. When the characteristic parameters are sequentially extracted from the frequency domain signals of the N frequency bands and the characteristic vectors are constructed, the characteristic matrix is constructed according to the characteristic vectors and is used as the input of the neural network model, the fault diagnosis result of the bearing is further obtained according to the output result of the neural network model, the whole bearing fault is judged without depending on the subjective judgment of testers, and therefore the bearing fault diagnosis accuracy can be improved.
Referring to fig. 3, an embodiment of the present invention provides a bearing fault diagnosis apparatus 30, including: an acquisition unit 301, a wavelet decomposition unit 302, a feature vector construction unit 303, a feature matrix construction unit 304, an input unit 305, and a fault diagnosis result acquisition unit 306, wherein:
the acquisition unit 301 is used for acquiring a vibration acceleration signal of the bearing in a working state;
a wavelet decomposition unit 302 for performing wavelet decomposition on the vibration acceleration signal;
a feature vector construction unit 303, configured to sequentially extract feature parameters from signals obtained by wavelet decomposition and construct feature vectors;
a feature matrix construction unit 304, configured to construct a feature matrix according to the feature vector;
an input unit 305, configured to input the feature matrix to a pre-trained neural network model;
a fault diagnosis result obtaining unit 306, configured to compare the output result of the neural network model with a preset target matrix, and obtain a fault diagnosis result of the bearing.
In a specific implementation, the fault diagnosis result of the bearing may include at least one of: whether the bearing is malfunctioning, the bearing element that is malfunctioning, and the severity of the malfunction.
In a specific implementation, the acquisition unit 301 may be configured to acquire a time-domain vibration acceleration signal of the bearing.
In a specific implementation, besides the acquisition unit 301, other functional units of the bearing fault diagnosis device 30 may be provided as a part of an online system or an offline system according to specific needs.
In a specific implementation, the wavelet decomposition unit 302 may be configured to select a wavelet mother function, where a waveform of the wavelet mother function is related to a waveform of a time-domain vibration acceleration signal of the bearing; performing wavelet decomposition on the wavelet mother function to obtain time domain signals of N frequency bands, wherein N is 2jAnd j is the level of wavelet decomposition.
In a specific implementation, the feature vector constructing unit 303 may be configured to demodulate the time domain signals of the N frequency bands and perform time-frequency conversion to obtain corresponding frequency domain signals of the N frequency bands; and sequentially extracting characteristic parameters from the frequency domain signals of the N frequency bands, and constructing a characteristic vector.
In a specific implementation, the feature matrix constructing unit 304 may be configured to construct a feature matrix according to the feature vector; the characteristic matrix is an M multiplied by N matrix, and M is the number of the characteristic vectors.
In a specific implementation, the target matrix may be an M × K matrix, where K is the number of components of the bearing that need to be diagnosed.
In a specific implementation, the bearing fault diagnosis apparatus 30 may further include: a denoising processing unit (not shown in fig. 3) for performing denoising processing on the vibration acceleration signal before the wavelet decomposition unit performs wavelet decomposition on the vibration acceleration signal.
In a specific implementation, the denoising processing unit may be configured to perform denoising processing on the vibration acceleration signal by using a wavelet threshold denoising method.
In particular implementations, the characteristic parameter may include at least one of: the energy of each wavelet decomposed signal, the crest factor of each wavelet decomposed signal, the standard deviation of each wavelet decomposed signal, the root mean square of each wavelet decomposed signal and the 90-quantile value of each wavelet decomposed signal.
In a specific implementation, the bearing fault diagnosis apparatus 30 may further include: a normalization processing unit (not shown in fig. 3) for normalizing the elements in the feature matrix before the input unit inputs the feature matrix to the pre-generated neural network model.
In a specific implementation, the neural network model may include: an input layer, a hidden layer, and an output layer; the input layer data of the neural network model may be the feature matrix, and the output layer data of the neural network model may be the target matrix.
In a specific implementation, the failure diagnosis result obtaining unit 306 may be configured to compare the output result of the neural network model with the target matrix, obtain an error value between the output result and the target matrix, and determine that the bearing component corresponding to the minimum error value of the target matrix has a failure.
Embodiments of the present invention further provide a computer-readable storage medium, on which computer instructions are stored, and when the computer instructions are executed, the steps of the bearing fault diagnosis method provided in the above embodiments of the present invention may be executed.
An embodiment of the present invention further provides an electronic device, where the electronic device includes a memory and a processor, where the memory stores computer instructions that can be executed on the processor, and the processor executes the computer instructions to perform the steps of the bearing fault diagnosis method provided in the above embodiment of the present invention.
Those skilled in the art will appreciate that all or part of the steps in the methods of the above embodiments may be implemented by instructing the relevant hardware through a program, which may be stored in a computer-readable storage medium, and the storage medium may include: ROM, RAM, magnetic or optical disks, and the like.
Although the present invention is disclosed above, the present invention is not limited thereto. Various changes and modifications may be effected therein by one skilled in the art without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (28)
1. A bearing fault diagnosis method, comprising:
collecting a vibration acceleration signal of a bearing in a working state;
carrying out wavelet decomposition on the vibration acceleration signal of the bearing;
sequentially extracting characteristic parameters from signals obtained by wavelet decomposition and constructing characteristic vectors;
constructing a feature matrix according to the feature vector;
inputting the feature matrix into a pre-trained neural network model;
and comparing the output result of the neural network model with a preset target matrix to obtain a fault diagnosis result of the bearing.
2. The bearing fault diagnosis method according to claim 1, wherein the fault diagnosis result of the bearing includes at least one of: whether the bearing is malfunctioning, the bearing element that is malfunctioning, and the severity of the malfunction.
3. The bearing fault diagnosis method according to claim 1 or 2, wherein the acquiring of the vibration acceleration signal of the bearing comprises:
and acquiring a time domain vibration acceleration signal of the bearing.
4. The bearing fault diagnosis method according to claim 1 or 2, wherein the performing wavelet decomposition on the vibration acceleration signal of the bearing comprises:
selecting a wavelet mother function, wherein the waveform of the wavelet mother function is related to the waveform of the time domain vibration acceleration signal of the bearing;
performing wavelet decomposition on the wavelet mother function to obtain time domain signals of N frequency bands, wherein N is 2jAnd j is the level of wavelet decomposition.
5. The bearing fault diagnosis method according to claim 4, wherein the sequentially extracting feature parameters from the signals obtained by wavelet decomposition and constructing feature vectors comprises:
demodulating the time domain signals of the N frequency bands and performing time-frequency conversion to obtain corresponding frequency domain signals of the N frequency bands;
and sequentially extracting characteristic parameters from the frequency domain signals of the N frequency bands, and constructing a characteristic vector.
6. The bearing fault diagnosis method according to claim 1 or 2, characterized in that said constructing a feature matrix from said feature vectors comprises:
constructing a feature matrix according to the feature vector; the characteristic matrix is an M multiplied by N matrix, and M is the number of the characteristic vectors.
7. A bearing fault diagnosis method according to claim 1 or 2, characterized in that the target matrix is an mxk matrix, K being the number of parts of the bearing to be diagnosed.
8. The bearing failure diagnosis method according to claim 1 or 2, characterized by, before the wavelet decomposition of the vibration acceleration signal, further comprising:
and carrying out noise reduction processing on the vibration acceleration signal.
9. The bearing fault diagnosis method according to claim 8, wherein the noise reduction processing of the vibration acceleration signal includes:
and denoising the vibration acceleration signal by adopting a wavelet threshold denoising method.
10. A bearing fault diagnosis method according to claim 1 or 2, characterized in that the characteristic parameters comprise at least one of:
the energy of each wavelet decomposed signal, the crest factor of each wavelet decomposed signal, the standard deviation of each wavelet decomposed signal, the root mean square of each wavelet decomposed signal and the 90-quantile value of each wavelet decomposed signal.
11. The bearing fault diagnosis method according to claim 1 or 2, characterized by, before inputting the feature matrix to a pre-generated neural network model, further comprising:
and carrying out normalization processing on elements in the feature matrix.
12. The bearing fault diagnosis method according to claim 1 or 2, characterized in that the neural network model comprises: an input layer, a hidden layer, and an output layer; and the input layer data of the neural network model is the characteristic matrix, and the output layer data of the neural network model is the target matrix.
13. The bearing fault diagnosis method according to claim 1 or 2, wherein the comparing the output result of the neural network model with a preset target matrix to obtain the fault diagnosis result of the bearing comprises:
comparing the output result of the neural network model with the target matrix, obtaining an error value between the output result and the target matrix, and judging that the bearing component corresponding to the minimum error value of the target matrix has a fault.
14. A bearing failure diagnosis device characterized by comprising:
the acquisition unit is used for acquiring a vibration acceleration signal of the bearing in a working state;
the wavelet decomposition unit is used for performing wavelet decomposition on the vibration acceleration signal of the bearing;
the characteristic vector construction unit is used for sequentially extracting characteristic parameters from signals obtained by wavelet decomposition and constructing characteristic vectors;
the characteristic matrix construction unit is used for constructing a characteristic matrix according to the characteristic vector;
the input unit is used for inputting the characteristic matrix to a pre-trained neural network model;
and the fault diagnosis result acquisition unit is used for comparing the output result of the neural network model with a preset target matrix to acquire the fault diagnosis result of the bearing.
15. The bearing malfunction diagnosis device according to claim 14, wherein the malfunction diagnosis result of the bearing includes at least one of: whether the bearing is malfunctioning, the bearing element that is malfunctioning, and the severity of the malfunction.
16. The bearing fault diagnosis device according to claim 14 or 15, characterized in that the acquisition unit is configured to acquire a time-domain vibration acceleration signal of the bearing.
17. The bearing fault diagnosis device according to claim 14 or 15, characterized in that the wavelet decomposition unit is configured to select a wavelet mother function, and the waveform of the wavelet mother function is related to the waveform of the time-domain vibration acceleration signal of the bearing; performing wavelet decomposition on the wavelet mother function to obtain time domain signals of N frequency bands, wherein N is 2jAnd j is the level of wavelet decomposition.
18. The bearing fault diagnosis device according to claim 17, wherein the eigenvector construction unit is configured to demodulate the time domain signals of the N frequency bands and perform time-frequency conversion to obtain frequency domain signals of the corresponding N frequency bands; and sequentially extracting characteristic parameters from the frequency domain signals of the N frequency bands, and constructing a characteristic vector.
19. The bearing failure diagnosis device according to claim 14 or 15, characterized by the feature matrix construction unit for constructing a feature matrix based on the feature vectors; the characteristic matrix is an M multiplied by N matrix, and M is the number of the characteristic vectors.
20. A bearing fault diagnosis apparatus according to claim 14 or 15, wherein the target matrix is an mxk matrix, K being the number of parts of the bearing to be diagnosed.
21. The bearing malfunction diagnosis device according to claim 14 or 15, further comprising: and the denoising processing unit is used for denoising the vibration acceleration signal before the wavelet decomposition unit performs wavelet decomposition on the vibration acceleration signal.
22. The bearing failure diagnostic device according to claim 21, wherein the noise reduction processing unit is configured to perform noise reduction processing on the vibration acceleration signal by using a wavelet threshold noise reduction method.
23. A bearing fault diagnosis apparatus according to claim 14 or 15, characterized in that the characteristic parameters comprise at least one of:
the energy of each wavelet decomposed signal, the crest factor of each wavelet decomposed signal, the standard deviation of each wavelet decomposed signal, the root mean square of each wavelet decomposed signal and the 90-quantile value of each wavelet decomposed signal.
24. The bearing malfunction diagnosis device according to claim 14 or 15, further comprising: and the normalization processing unit is used for normalizing the elements in the characteristic matrix before the input unit inputs the characteristic matrix to a pre-generated neural network model.
25. The bearing fault diagnosis apparatus according to claim 14 or 15, characterized in that the neural network model includes: an input layer, a hidden layer, and an output layer; and the input layer data of the neural network model is the characteristic matrix, and the output layer data of the neural network model is the target matrix.
26. The bearing failure diagnosis device according to claim 14 or 15, wherein the failure diagnosis result obtaining unit is configured to compare an output result of the neural network model with the target matrix, obtain an error value of the output result with the target matrix, and determine that the bearing component corresponding to the minimum error value of the target matrix has a failure.
27. A computer readable storage medium having computer instructions stored thereon, wherein the computer instructions when executed perform the steps of the bearing fault diagnosis method of claim 1 or 2.
28. An electronic device comprising a memory and a processor, the memory having stored thereon computer instructions executable on the processor, wherein the processor, when executing the computer instructions, performs the steps of the bearing fault diagnosis method of claim 1 or 2.
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2017/103304 WO2019061006A1 (en) | 2017-09-26 | 2017-09-26 | Bearing failure diagnosis method and device, readable storage medium, and electronic device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111094927A true CN111094927A (en) | 2020-05-01 |
Family
ID=65900314
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201780094590.0A Pending CN111094927A (en) | 2017-09-26 | 2017-09-26 | Bearing fault diagnosis method and device, readable storage medium and electronic equipment |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN111094927A (en) |
| WO (1) | WO2019061006A1 (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111680665A (en) * | 2020-06-28 | 2020-09-18 | 湖南大学 | Motor mechanical fault diagnosis method based on data driving and adopting current signals |
| CN112067289A (en) * | 2020-08-21 | 2020-12-11 | 天津电气科学研究院有限公司 | Motor shaft and transmission shaft abnormal vibration early warning algorithm based on neural network |
| CN112146880A (en) * | 2020-09-17 | 2020-12-29 | 天津大学 | Intelligent diagnosis method for internal structure faults of rolling bearing at different rotating speeds |
| CN113642433A (en) * | 2021-07-30 | 2021-11-12 | 深圳市中金岭南有色金属股份有限公司凡口铅锌矿 | Method and device for determining working state of elevator, terminal equipment and medium |
| CN113705096A (en) * | 2021-08-27 | 2021-11-26 | 北京博华信智科技股份有限公司 | Small sample deep learning-based impact fault diagnosis |
| CN114088401A (en) * | 2021-11-03 | 2022-02-25 | 宁波坤博测控科技有限公司 | Fault analysis method and device for rolling bearing of wind driven generator |
| CN114104224A (en) * | 2021-11-15 | 2022-03-01 | 中国船舶重工集团公司第七一一研究所 | Device management method, device, electronic device and computer-readable storage medium |
| CN114659785A (en) * | 2021-12-27 | 2022-06-24 | 三一重能股份有限公司 | Fault detection method and device for transmission chain of wind driven generator |
| CN114659785B (en) * | 2021-12-27 | 2024-07-26 | 三一重能股份有限公司 | Fault detection method and device for wind driven generator transmission chain |
Families Citing this family (38)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111832353B (en) * | 2019-04-19 | 2023-09-05 | 中国科学院沈阳自动化研究所 | Steam turbine rotor fault diagnosis method based on EMD and BA optimization SVM |
| CN110222390B (en) * | 2019-05-24 | 2023-07-18 | 武汉工程大学 | Gear crack identification method based on wavelet neural network |
| CN110188485B (en) * | 2019-06-03 | 2020-08-25 | 安徽理工大学 | Self-adaptive obtaining method for dynamic performance degradation trend of rolling bearing |
| CN110046476B (en) * | 2019-06-05 | 2020-10-16 | 厦门大学 | Ternary binary fractal wavelet sparse diagnosis method for rolling bearing faults |
| CN110146156B (en) * | 2019-06-27 | 2021-02-09 | 南昌航空大学 | Denoising method for fault vibration signal of aircraft engine rotor system |
| CN110672323B (en) * | 2019-09-02 | 2021-03-26 | 佛山科学技术学院 | Bearing health state assessment method and device based on neural network |
| CN110795843B (en) * | 2019-10-24 | 2024-03-29 | 北京建筑大学 | Method and device for identifying faults of rolling bearing |
| CN110826607A (en) * | 2019-10-24 | 2020-02-21 | 北京建筑大学 | Fault detection method and device for rolling bearing |
| CN110781854B (en) * | 2019-11-04 | 2022-07-26 | 上海联影智能医疗科技有限公司 | Training method of fault detection model and fault detection method of electromechanical equipment |
| CN110866519B (en) * | 2019-12-24 | 2023-04-28 | 安徽工业大学 | Rolling bearing fault diagnosis method based on Fourier decomposition and multiscale permutation entropy deviation value |
| CN110991564B (en) * | 2019-12-24 | 2023-05-26 | 安徽工业大学 | Variable working condition bearing fault diagnosis method based on multiscale dispersion entropy deviation mean value and nonlinear mode decomposition |
| CN111222495B (en) * | 2020-02-17 | 2023-05-12 | 济南大学 | Rolling bearing fault diagnosis method based on wavelet decomposition and improved Manhattan distance |
| CN111325159B (en) * | 2020-02-25 | 2023-09-08 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Fault diagnosis method, device, computer equipment and storage medium |
| CN111275024B (en) * | 2020-03-22 | 2023-03-21 | 东北电力大学 | Fault diagnosis method for rolling bearing of variable-working-condition fan driven by limited data containing noise label |
| CN111625762B (en) * | 2020-04-29 | 2023-07-25 | 广东省能源集团贵州有限公司 | Fan fault diagnosis method |
| EP3929554A1 (en) * | 2020-06-26 | 2021-12-29 | Siemens Aktiengesellschaft | Improved fault detection in machines using ai |
| CN111797789B (en) * | 2020-07-10 | 2024-02-06 | 合肥工业大学 | Time-frequency ridge line extraction method and device for fault feature interpretation |
| CN111914703A (en) * | 2020-07-20 | 2020-11-10 | 哈尔滨工业大学 | Mechanical rotating part fault diagnosis method based on wavelet transformation and transfer learning GoogLeNet |
| CN112304611A (en) * | 2020-09-14 | 2021-02-02 | 江苏大学 | Deep learning-based bearing fault diagnosis method |
| CN112257530B (en) * | 2020-10-13 | 2023-10-20 | 西安电子科技大学 | Rolling bearing fault diagnosis method based on blind signal separation and support vector machine |
| CN112284735B (en) * | 2020-10-21 | 2022-07-15 | 兰州理工大学 | Multi-sensor rolling bearing fault diagnosis based on one-dimensional convolution and dynamic routing |
| CN112329650A (en) * | 2020-11-09 | 2021-02-05 | 中国兵器工业集团第二一四研究所苏州研发中心 | Accelerometer fault diagnosis method based on convolutional neural network |
| CN112595515A (en) * | 2020-12-04 | 2021-04-02 | 中国船舶工业综合技术经济研究院 | Power shafting bearing fault detection method and system |
| CN112465052B (en) * | 2020-12-07 | 2023-04-07 | 重庆忽米网络科技有限公司 | Equipment fault diagnosis report generation method and system based on convolutional neural network |
| CN112561306B (en) * | 2020-12-11 | 2023-12-08 | 领伟创新智能***(浙江)有限公司 | Rolling bearing health state evaluation method based on Hankel matrix |
| CN112629863B (en) * | 2020-12-31 | 2022-03-01 | 苏州大学 | Bearing fault diagnosis method for dynamic joint distribution alignment network under variable working conditions |
| CN112766219A (en) * | 2021-01-31 | 2021-05-07 | 石家庄铁道大学 | Bearing health assessment method and device based on depth separable convolution self-encoder |
| CN112857805B (en) * | 2021-03-13 | 2022-05-31 | 宁波大学科学技术学院 | Rolling bearing fault detection method based on graph similarity feature extraction |
| CN113532835A (en) * | 2021-08-18 | 2021-10-22 | 中国国家铁路集团有限公司 | Railway contact net hard spot diagnosis method and device |
| CN113821888B (en) * | 2021-09-23 | 2024-02-27 | 西安热工研究院有限公司 | Vibration data fault diagnosis method based on periodic impact feature extraction and echo state network |
| CN114441174B (en) * | 2022-02-09 | 2024-07-09 | 上海电气集团股份有限公司 | Method, system, equipment and medium for diagnosing composite fault of rolling bearing |
| CN114781427B (en) * | 2022-03-17 | 2024-05-14 | 华电河南新能源发电有限公司 | Wind generating set antifriction bearing fault diagnosis system based on information fusion technology |
| CN114757060B (en) * | 2022-06-15 | 2022-12-20 | 石家庄铁道大学 | Bearing fault diagnosis method based on morphological function construction scale space |
| CN115081491B (en) * | 2022-08-18 | 2022-11-08 | 深圳市乐闵科技有限公司 | Running state fault diagnosis method for sanitary towel production equipment |
| CN116183231B (en) * | 2023-04-28 | 2023-07-11 | 安徽大学 | Bearing fault diagnosis method based on digital twin |
| CN116296377B (en) * | 2023-05-10 | 2023-08-01 | 北京奔驰汽车有限公司 | Industrial robot RV reducer fault prediction method and device |
| CN116536850A (en) * | 2023-05-11 | 2023-08-04 | 郑州豫力无纺布有限公司 | High-speed hot rolling mill for non-woven fabrics |
| CN116541713B (en) * | 2023-06-27 | 2023-12-12 | 清华大学 | Bearing fault diagnosis model training method based on local time-frequency characteristic transfer learning |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008292288A (en) * | 2007-05-24 | 2008-12-04 | Mitsubishi Electric Engineering Co Ltd | Bearing diagnostic device for reduction gear |
| KR20120121621A (en) * | 2011-04-27 | 2012-11-06 | 경희대학교 산학협력단 | Diagnostic apparatus for vehicle, diagnostic method for vehicle and recording medium of the same diagnostic method |
| CN102829974A (en) * | 2012-08-07 | 2012-12-19 | 北京交通大学 | LMD (local mean decomposition) and PCA (principal component analysis) based rolling bearing state identification method |
| CN103076177A (en) * | 2013-01-16 | 2013-05-01 | 昆明理工大学 | Rolling bearing fault detection method based on vibration detection |
| CN103914617A (en) * | 2014-03-25 | 2014-07-09 | 北京交通大学 | Fault diagnosis method for subway vehicle bogie bearings |
| CN104807639A (en) * | 2015-04-23 | 2015-07-29 | 广西大学 | Fault diagnosis method and device for rolling bearing of running gear of locomotive |
| CN106650071A (en) * | 2016-12-12 | 2017-05-10 | 中国航空工业集团公司上海航空测控技术研究所 | Intelligent fault diagnosis method for rolling bearing |
-
2017
- 2017-09-26 WO PCT/CN2017/103304 patent/WO2019061006A1/en active Application Filing
- 2017-09-26 CN CN201780094590.0A patent/CN111094927A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008292288A (en) * | 2007-05-24 | 2008-12-04 | Mitsubishi Electric Engineering Co Ltd | Bearing diagnostic device for reduction gear |
| KR20120121621A (en) * | 2011-04-27 | 2012-11-06 | 경희대학교 산학협력단 | Diagnostic apparatus for vehicle, diagnostic method for vehicle and recording medium of the same diagnostic method |
| CN102829974A (en) * | 2012-08-07 | 2012-12-19 | 北京交通大学 | LMD (local mean decomposition) and PCA (principal component analysis) based rolling bearing state identification method |
| CN103076177A (en) * | 2013-01-16 | 2013-05-01 | 昆明理工大学 | Rolling bearing fault detection method based on vibration detection |
| CN103914617A (en) * | 2014-03-25 | 2014-07-09 | 北京交通大学 | Fault diagnosis method for subway vehicle bogie bearings |
| CN104807639A (en) * | 2015-04-23 | 2015-07-29 | 广西大学 | Fault diagnosis method and device for rolling bearing of running gear of locomotive |
| CN106650071A (en) * | 2016-12-12 | 2017-05-10 | 中国航空工业集团公司上海航空测控技术研究所 | Intelligent fault diagnosis method for rolling bearing |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111680665A (en) * | 2020-06-28 | 2020-09-18 | 湖南大学 | Motor mechanical fault diagnosis method based on data driving and adopting current signals |
| CN112067289A (en) * | 2020-08-21 | 2020-12-11 | 天津电气科学研究院有限公司 | Motor shaft and transmission shaft abnormal vibration early warning algorithm based on neural network |
| CN112146880A (en) * | 2020-09-17 | 2020-12-29 | 天津大学 | Intelligent diagnosis method for internal structure faults of rolling bearing at different rotating speeds |
| CN112146880B (en) * | 2020-09-17 | 2022-03-29 | 天津大学 | Intelligent diagnosis method for internal structure faults of rolling bearing at different rotating speeds |
| CN113642433A (en) * | 2021-07-30 | 2021-11-12 | 深圳市中金岭南有色金属股份有限公司凡口铅锌矿 | Method and device for determining working state of elevator, terminal equipment and medium |
| CN113642433B (en) * | 2021-07-30 | 2024-04-02 | 深圳市中金岭南有色金属股份有限公司凡口铅锌矿 | Method, device, terminal equipment and medium for determining working state of elevator |
| CN113705096A (en) * | 2021-08-27 | 2021-11-26 | 北京博华信智科技股份有限公司 | Small sample deep learning-based impact fault diagnosis |
| CN114088401A (en) * | 2021-11-03 | 2022-02-25 | 宁波坤博测控科技有限公司 | Fault analysis method and device for rolling bearing of wind driven generator |
| CN114104224A (en) * | 2021-11-15 | 2022-03-01 | 中国船舶重工集团公司第七一一研究所 | Device management method, device, electronic device and computer-readable storage medium |
| CN114659785A (en) * | 2021-12-27 | 2022-06-24 | 三一重能股份有限公司 | Fault detection method and device for transmission chain of wind driven generator |
| CN114659785B (en) * | 2021-12-27 | 2024-07-26 | 三一重能股份有限公司 | Fault detection method and device for wind driven generator transmission chain |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2019061006A1 (en) | 2019-04-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111094927A (en) | Bearing fault diagnosis method and device, readable storage medium and electronic equipment | |
| CN109977920B (en) | Water turbine set fault diagnosis method based on time-frequency spectrogram and convolutional neural network | |
| CN107192553B (en) | Gear-box combined failure diagnostic method based on blind source separating | |
| CN102721545B (en) | Rolling bearing failure diagnostic method based on multi-characteristic parameter | |
| CN102661783B (en) | Characteristic extracting method for prediction of rotating mechanical failure trend | |
| CN107147397B (en) | Wearable device-oriented rapid compressed sensing reconstruction method | |
| He et al. | Sparse signal reconstruction based on time–frequency manifold for rolling element bearing fault signature enhancement | |
| CN110795843B (en) | Method and device for identifying faults of rolling bearing | |
| CN109827776B (en) | Bearing fault detection method and system | |
| Yi et al. | Mechanical compound faults extraction based on improved frequency domain blind deconvolution algorithm | |
| Pandya et al. | ANN based fault diagnosis of rolling element bearing using time-frequency domain feature | |
| CN114636556A (en) | Method for judging bearing fault based on CEEMDAN decomposition, electronic device and storage medium | |
| CN115730199B (en) | Rolling bearing vibration signal noise reduction and fault feature extraction method and system | |
| Huang et al. | A Fault Diagnosis Approach for Rolling Bearing Based on Wavelet Packet Decomposition and GMM-HMM. | |
| Zhang et al. | Sparsity-aware tight frame learning with adaptive subspace recognition for multiple fault diagnosis | |
| CN106502850B (en) | A kind of the multiple faults discrimination method and device of sparse tight frame dictionary learning model | |
| CN106301610A (en) | The adaptive failure detection of a kind of superhet and diagnostic method and device | |
| CN104075889B (en) | Null space bearing composite fault diagnosis method based on fault features | |
| CN114263621A (en) | Test method and system for diagnosing and simulating cavitation fault of centrifugal pump | |
| CN114264478A (en) | Diesel engine crankshaft bearing wear degree prediction method and system | |
| CN110222390B (en) | Gear crack identification method based on wavelet neural network | |
| CN110147637B (en) | Rub-impact fault diagnosis method based on wavelet and harmonic component greedy sparse identification | |
| CN114441174B (en) | Method, system, equipment and medium for diagnosing composite fault of rolling bearing | |
| CN116086807A (en) | Fault judging method for rolling bearing and related product | |
| CN112595515A (en) | Power shafting bearing fault detection method and system |
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 |