CN116484176A - Bearing fault diagnosis method, system and storage medium based on ultra-wavelet - Google Patents
Bearing fault diagnosis method, system and storage medium based on ultra-wavelet Download PDFInfo
- Publication number
- CN116484176A CN116484176A CN202310462774.5A CN202310462774A CN116484176A CN 116484176 A CN116484176 A CN 116484176A CN 202310462774 A CN202310462774 A CN 202310462774A CN 116484176 A CN116484176 A CN 116484176A
- Authority
- CN
- China
- Prior art keywords
- wavelet
- bearing
- ultra
- fault
- vibration signal
- 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 74
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000003860 storage Methods 0.000 title claims abstract description 17
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 56
- 230000004927 fusion Effects 0.000 claims abstract description 47
- 230000006870 function Effects 0.000 claims abstract description 45
- 230000009467 reduction Effects 0.000 claims abstract description 23
- 238000012549 training Methods 0.000 claims description 54
- 230000003044 adaptive effect Effects 0.000 claims description 22
- 230000009466 transformation Effects 0.000 claims description 22
- 238000012360 testing method Methods 0.000 claims description 21
- 238000012795 verification Methods 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000000354 decomposition reaction Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 7
- 238000004091 panning Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 abstract description 7
- 238000000605 extraction Methods 0.000 abstract description 5
- 238000004458 analytical method Methods 0.000 description 20
- 238000005070 sampling Methods 0.000 description 14
- 238000005096 rolling process Methods 0.000 description 13
- 230000008901 benefit Effects 0.000 description 11
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 8
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000009826 distribution Methods 0.000 description 6
- 230000036039 immunity Effects 0.000 description 6
- 238000010606 normalization Methods 0.000 description 6
- 238000011176 pooling Methods 0.000 description 6
- 238000002474 experimental method Methods 0.000 description 5
- 230000000295 complement effect Effects 0.000 description 4
- 230000002708 enhancing effect Effects 0.000 description 4
- 238000004088 simulation Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 108010076504 Protein Sorting Signals Proteins 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000007797 corrosion Effects 0.000 description 2
- 238000005260 corrosion Methods 0.000 description 2
- 238000005520 cutting process Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000003754 machining Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
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
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/70—Wind energy
- Y02E10/72—Wind turbines with rotation axis in wind direction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (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
The invention discloses a bearing fault diagnosis method, a system and a storage medium based on ultra-wavelet, wherein the method comprises the following steps: acquiring a bearing vibration signal; inputting the bearing vibration signal into a preset ultra-wavelet convolutional neural network model, and analyzing the bearing vibration signal through an ultra-wavelet block to obtain bearing fault characteristics of multiple scales; weighting and fusing the bearing fault characteristics of the multiple scales to obtain fused fault characteristics; and analyzing according to the fusion fault characteristics to obtain bearing fault diagnosis information. According to the invention, the vibration signal of the fan gear box bearing is processed through the SWCNN, the bearing vibration signal containing noise is processed through the ultra-wavelet block, and the noise reduction processing and the feature extraction are respectively carried out by using 3 different wavelet basis functions, so that the noise resistance of the SWCNN network is enhanced. In addition, the weight fusion layer of the ultra-wavelet block can adaptively determine fusion weights, so that wavelet channels with obvious fault characteristics are enhanced, key characteristics are highlighted, and diagnosis accuracy is improved.
Description
Technical Field
The present application relates to the field of data processing and data transmission, and more particularly, to a blower gearbox bearing fault diagnosis method, system and storage medium based on ultra-wavelet.
Background
The rolling bearing is widely applied to a fan gear box and is one of key parts for affecting the normal operation of the fan gear box. However, the rolling bearing is in a state of high-speed rotation and heavy load for a long time, and 45% -55% of faults of rotating equipment are caused by bearing damage. The failure of the rolling bearing will lead to a long-time stoppage of the equipment, leading to an increase in production and manufacturing costs and even safety accidents. The running state of the rolling bearing of the fan gear box is monitored and diagnosed, so that problems can be found out in time, faults can be removed, further, safety accidents are reduced, and the method has important social significance and economic value. Therefore, it is important to research and develop an effective fan gearbox bearing fault diagnosis method.
With the development of artificial intelligence technology, bearing fault diagnosis research based on deep learning is widely focused by experts and scholars. Among them, CNN is a typical deep learning method, which has been successful in various fields of image classification, object detection, natural language processing, voice recognition, and the like. However, due to the severe working environment and large background noise of the rolling bearing, the vibration signal of the bearing is interfered by noise, so that the effective fault characteristics of the CNN model are difficult to extract, and the fault diagnosis accuracy of the CNN model is lowered.
Therefore, the prior art has defects, and improvement is needed.
Disclosure of Invention
In view of the above problems, an object of the present invention is to provide a blower gearbox bearing fault diagnosis method, system and storage medium based on ultra-wavelet, which can realize diagnosis of bearing faults under the condition of strong noise.
The first aspect of the invention provides a bearing fault diagnosis method based on ultra-wavelet, comprising the following steps:
acquiring a fan gear box bearing vibration signal;
inputting the bearing vibration signal into a preset ultra-wavelet convolutional neural network model, and analyzing the bearing vibration signal through an ultra-wavelet block to obtain bearing fault characteristics of multiple scales;
weighting and fusing the bearing fault characteristics of the multiple scales to obtain fused fault characteristics;
and analyzing according to the fusion fault characteristics to obtain bearing fault diagnosis information.
In this scheme, still include:
establishing an initial ultra-wavelet convolutional neural network model;
acquiring vibration signal sample data;
dividing the vibration signal sample data into a training set, a verification set and a test set according to a preset proportion;
and respectively inputting the training set, the verification set and the test set into the initial ultra-wavelet convolutional neural network model for training and diagnosis to obtain a preset ultra-wavelet convolutional neural network model.
In this scheme, after will bearing vibration signal input to the ultra wavelet convolutional neural network model of predetermineeing, still include:
acquiring a plurality of wavelet base data;
analyzing the wavelet base data to obtain a plurality of self-adaptive wavelets;
the calculation method of the adaptive wavelet is expressed as the following formula:
wherein,,is the nth wavelet basis function, a n Is the scale factor of the wavelet basis function, b n Is a panning factor of the wavelet basis function, w n Is the combination coefficient of the ultra-wavelet, k represents the kth wavelet base function.
In this scheme, through ultra-wavelet piece with bearing vibration signal analysis obtains the bearing fault characteristic of a plurality of scales, includes:
the ultra-wavelet block carries out wavelet transformation on the bearing vibration signals through the plurality of self-adaptive wavelets respectively, so that the ultra-wavelet block carries out decomposition and noise reduction on the bearing vibration signals through wavelet transformation to obtain noise reduction vibration signals;
extracting features of the noise reduction vibration signals through the plurality of adaptive wavelets to obtain bearing fault features with a plurality of scales;
the wavelet transform is formulated as:
where ψ represents the mother wavelet, t is time, a is a scale factor inversely proportional to frequency, and b is a shift factor.
In this scheme, the bearing fault characteristics of a plurality of scales are weighted and fused to obtain a fused fault characteristic, including:
multiplying the bearing fault characteristics of the multiple scales by corresponding influence weights respectively to obtain weight scores of the multiple fault characteristics;
selecting a plurality of key fault characteristics according to the weight scores of the plurality of fault characteristics;
and carrying out feature fusion on the plurality of key fault features to obtain fusion fault features.
In this scheme, the analysis is performed according to the fusion fault characteristics to obtain bearing fault diagnosis information, including:
inputting the fused fault signature to a plurality of convolution layers;
the plurality of convolution layers process the fusion fault characteristics according to a preset method and send the processed fault characteristics to a full connection layer;
and the full-connection layer integrates and classifies the processed fault characteristics to obtain bearing fault diagnosis information.
The invention provides a blower gear box bearing fault diagnosis system based on ultra-wavelet, which comprises a memory and a processor, wherein the memory comprises a blower gear box bearing fault diagnosis method program based on ultra-wavelet, and the blower gear box bearing fault diagnosis method program based on ultra-wavelet realizes the following steps when being executed by the processor:
Acquiring a fan gear box bearing vibration signal;
inputting the bearing vibration signal into a preset ultra-wavelet convolutional neural network model, and analyzing the bearing vibration signal through an ultra-wavelet block to obtain bearing fault characteristics of multiple scales;
weighting and fusing the bearing fault characteristics of the multiple scales to obtain fused fault characteristics;
and analyzing according to the fusion fault characteristics to obtain bearing fault diagnosis information.
In this scheme, still include:
establishing an initial ultra-wavelet convolutional neural network model;
acquiring fan gear box vibration signal sample data;
dividing the vibration signal sample data into a training set, a verification set and a test set according to a preset proportion;
and respectively inputting the training set, the verification set and the test set into the initial ultra-wavelet convolutional neural network model for training and diagnosis to obtain a preset ultra-wavelet convolutional neural network model.
In this scheme, after will bearing vibration signal input to the ultra wavelet convolutional neural network model of predetermineeing, still include:
acquiring a plurality of wavelet base data;
analyzing the wavelet base data to obtain a plurality of self-adaptive wavelets;
the calculation method of the adaptive wavelet is expressed as the following formula:
Wherein,,is the nth wavelet basis function, a n Is the scale factor of the wavelet basis function, b n Is a panning factor of the wavelet basis function, w n Is the combination coefficient of the ultra-wavelet, k represents the kth wavelet base function.
In this scheme, through ultra-wavelet piece with bearing vibration signal analysis obtains the bearing fault characteristic of a plurality of scales, includes:
the ultra-wavelet block carries out wavelet transformation on the bearing vibration signals through the plurality of self-adaptive wavelets respectively, so that the ultra-wavelet block carries out decomposition and noise reduction on the bearing vibration signals through wavelet transformation to obtain noise reduction vibration signals;
extracting features of the noise reduction vibration signals through the plurality of adaptive wavelets to obtain bearing fault features with a plurality of scales;
the wavelet transform is formulated as:
where ψ represents the mother wavelet, t is time, a is a scale factor inversely proportional to frequency, and b is a shift factor.
In this scheme, the bearing fault characteristics of a plurality of scales are weighted and fused to obtain a fused fault characteristic, including:
multiplying the bearing fault characteristics of the multiple scales by corresponding influence weights respectively to obtain weight scores of the multiple fault characteristics;
Selecting a plurality of key fault characteristics according to the weight scores of the plurality of fault characteristics;
and carrying out feature fusion on the plurality of key fault features to obtain fusion fault features.
In this scheme, the analysis is performed according to the fusion fault characteristics to obtain bearing fault diagnosis information, including:
inputting the fused fault signature to a plurality of convolution layers;
the plurality of convolution layers process the fusion fault characteristics according to a preset method and send the processed fault characteristics to a full connection layer;
and the full-connection layer integrates and classifies the processed fault characteristics to obtain bearing fault diagnosis information.
A third aspect of the present invention provides a computer-readable storage medium, in which a blower gearbox bearing fault diagnosis method program based on ultra-wavelets is included, which when executed by a processor, implements the steps of a blower gearbox bearing fault diagnosis method based on ultra-wavelets as described in any one of the above.
The invention discloses a fan gear box bearing fault diagnosis method, a system and a storage medium based on ultra-wavelet, wherein the method comprises the following steps: acquiring a bearing vibration signal; inputting the bearing vibration signal into a preset ultra-wavelet convolutional neural network model, and analyzing the bearing vibration signal through an ultra-wavelet block to obtain bearing fault characteristics of multiple scales; weighting and fusing the bearing fault characteristics of the multiple scales to obtain fused fault characteristics; and analyzing according to the fusion fault characteristics to obtain bearing fault diagnosis information. According to the invention, the bearing vibration signal is processed through the SWCNN, the bearing vibration signal containing noise is processed through the ultra-wavelet block, and the noise reduction processing and the feature extraction are respectively carried out by using 3 different wavelet basis functions, so that the noise resistance of the SWCNN network is enhanced. In addition, the weight fusion layer of the ultra-wavelet block can adaptively determine fusion weights, so that wavelet channels with obvious fault characteristics are enhanced, key characteristics are highlighted, and the diagnosis accuracy of SWCNN is improved.
Drawings
FIG. 1 shows a flow chart of a blower gearbox bearing fault diagnosis method based on ultra-wavelet of the present invention;
FIG. 2 is a flow chart of a method for training a preset ultra-wavelet convolutional neural network model of the present invention;
FIG. 3 is a flow chart showing a method of acquiring bearing fault diagnosis information according to the present invention;
FIG. 4 shows a block diagram of an ultra-wavelet based bearing failure diagnosis system of the present invention;
FIG. 5 shows an illustration of a bearing data acquisition system of the present invention;
fig. 6 shows a training flow chart of an inventive ultra-wavelet convolutional neural network.
Detailed Description
In order that the above-recited objects, features and advantages of the present invention will be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description. It should be noted that, in the case of no conflict, the embodiments of the present application and the features in the embodiments may be combined with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those described herein, and therefore the scope of the present invention is not limited to the specific embodiments disclosed below.
FIG. 1 shows a flow chart of a blower gearbox bearing fault diagnosis method based on ultra-wavelets.
As shown in fig. 1, the invention discloses a blower gear box bearing fault diagnosis method based on ultra-wavelet, which comprises the following steps:
s102, obtaining a vibration signal of a bearing of a fan gear box;
s104, inputting the bearing vibration signal into a preset ultra-wavelet convolutional neural network model, and analyzing the bearing vibration signal through an ultra-wavelet block to obtain bearing fault characteristics of multiple scales;
s106, carrying out weighted fusion on the bearing fault characteristics of the multiple scales to obtain fusion fault characteristics;
s108, analyzing according to the fusion fault characteristics to obtain bearing fault diagnosis information.
According to the embodiment of the invention, the preset ultra-wavelet convolutional neural network (SWCNN) is established according to the wavelet transformation principle. The SWCNN takes the ultra-wavelet convolution block as a first layer of the CNN network, so that the noise of the bearing vibration signal is reduced, and the key fault characteristics are enhanced. The scheme selects Morlet wavelet, laplace wavelet and Mexhat wavelet to form wavelet core of ultra-wavelet block. The ultra-wavelet block adaptively selects and constructs a wavelet basis which can effectively represent a bearing fault signal from an original bearing vibration signal, and simultaneously adopts a plurality of wavelet basis functions to carry out wavelet transformation, so that bearing fault characteristics are highlighted on different scales, and multi-scale fault characteristics are provided for a CNN model. In addition, the ultra-wavelet block can decompose key characteristic information from noise through the characteristic of wavelet transformation multi-resolution analysis, so that the noise immunity of the CNN model is improved.
SWCNN adopts Adaptive Wavelet Transform (AWT) and adopts Laplace wavelet, morlet wavelet and Mexh wavelet to form adaptive wavelet, so that the advantages of different wavelet bases can be fused. In addition, SWCNN does not need to manually select a wavelet base, and the self-adaptive wavelet is optimized by CNN on wavelet parameters and combination coefficients, so that a group of self-adaptive wavelet filters with good noise reduction performance are learned. The present protocol analyzes SWCNN through two different bearing datasets. The result shows that the SWCNN can well retain the original information, and fully considers the internal relation of the sequence data, thereby realizing fault diagnosis under the condition of strong noise.
FIG. 2 shows a flow chart of a method for training a preset ultra-wavelet convolutional neural network model of the present invention.
As shown in fig. 2, according to an embodiment of the present invention, further includes:
s202, an initial ultra-wavelet convolutional neural network model is established;
s204, obtaining vibration signal sample data;
s206, dividing the vibration signal sample data into a training set, a verification set and a test set according to a preset proportion;
s208, respectively inputting the training set, the verification set and the test set into the initial ultra-wavelet convolutional neural network model for training and diagnosis to obtain a preset ultra-wavelet convolutional neural network model.
It should be noted that the bearing data set of the experiment in the scheme comes from the fan gear box simulation bearing fault simulation comprehensive test bed. The experimental device is shown in fig. 5, and the experimental bench mainly comprises a driving motor, a transmission part, a supporting part, a controller and the like. In the rolling bearing fault experiment, four signals are collected, and a normal bearing, an inner ring fault bearing, an outer ring fault bearing and a rolling body fault bearing are obtained. The sampling frequency is 8192Hz, the acquisition time is 120s, and the length of the sampling signal sequence of each fault condition is 983040.
Faults such as bearing cracks, pitting corrosion and the like in actual working conditions are simulated, faults with different depths are implanted into an inner ring, an outer ring and a rolling body of the rolling bearing in a mode of linear cutting and groove machining, and the depths are 0.4mm, 0.8mm and 1.2mm respectively. Healthy bearings can be considered a special failure mode, so there are ten types of bearing failure in the experimental dataset.
The training flow of the SWCNN model is shown in FIG. 6, and comprises 3 steps of data set division, training of the model and model diagnosis, wherein the specific steps are as follows:
(1) Dividing the original data, namely the vibration signal sample data, into three data sets according to a preset proportion: training set, verification set and test set, preset proportion is 8:1:1. wherein the data of the training set is raw noise-free data so that SWCNN can learn the appropriate wavelet basis. And the verification set and the test set are added with noise with a certain signal to noise ratio so as to judge whether the trained SWCNN model can resist noise interference and diagnose bearing faults.
(2) When training the model, firstly, initializing network parameters of SWCNN. And then inputting a training set to train parameters of each layer of SWCNN, so that the SWCNN can fully learn the characteristics of the original data. And finally, checking the trained SWCNN model by using a verification set containing different noises, and judging whether the target accuracy is reached. Repeating the steps for a plurality of times, and selecting an optimal model as a diagnosis model.
(3) In the model diagnosis, the optimal model in the step (2) is firstly taken as a diagnosis model. The test set is then input into a diagnostic model. Finally, the result of the diagnostic model is used as a judgment standard for evaluating SWCNN performance.
According to an embodiment of the present invention, after the bearing vibration signal is input to a preset ultra-wavelet convolutional neural network model, the method further includes:
acquiring wavelet database data;
analyzing the wavelet base data to obtain a plurality of self-adaptive wavelets;
the calculation method of the adaptive wavelet is expressed as the following formula:
wherein,,is the nth wavelet basis function, a n Is the scale factor of the wavelet basis function, b n Is a panning factor of the wavelet basis function, w n Is the combination coefficient of the ultra-wavelet, k represents the kth wavelet base function.
It should be noted that the wavelet transform can decompose the bearing vibration signal to reduce the noise and interference of the bearing vibration signal, and can obtain more information by utilizing the characteristics of multi-resolution analysis. To take advantage of the characteristics of wavelet transform multi-resolution analysis, the present solution proposes a Super Wavelet Block (SWB). SWB consists of two convolutional layers. The first convolution layer is a wavelet transform layer and is formed by a plurality of different wavelet channels in parallel, and each wavelet channel represents a wavelet base. The second convolution layer is a weighted fusion layer, which is a 1×1 standard convolution layer, used to perform weighted fusion on the characteristics of different wavelet channel outputs.
The SWB principle is that fault characteristics are extracted from bearing vibration signals by utilizing adaptive wavelets, the adaptive wavelets adopt a multi-wavelet-base complementary mechanism, complementary characteristics are obtained by utilizing different wavelet bases, different weights are given to the characteristics, and the bearing fault characteristics are fully extracted. Furthermore, the adaptive wavelet satisfies the wavelet's tolerability condition, i.e. it is regarded as a wavelet itself. Therefore, SWB can merge the advantages of different wavelet bases, obtain more fault characteristic information, and learn new wavelet bases.
The SWB has two main advantages, namely the SWB performs feature extraction through a plurality of wavelet basis functions so as to match different fault features in the vibration signal, obtain more useful information and further increase the noise immunity of the CNN model. The second advantage is that SWB can directly input original vibration signal, well keep original information, reduce the original vibration signal information loss that adopts denoising preliminary treatment to cause.
According to an embodiment of the present invention, the analyzing the bearing vibration signal by the ultra-wavelet block to obtain bearing fault characteristics with multiple dimensions includes:
the ultra-wavelet block carries out wavelet transformation on the bearing vibration signals through the plurality of self-adaptive wavelets respectively, so that the ultra-wavelet block carries out decomposition and noise reduction on the bearing vibration signals through wavelet transformation to obtain noise reduction vibration signals;
extracting features of the noise reduction vibration signals through the plurality of adaptive wavelets to obtain bearing fault features with a plurality of scales;
the wavelet transform is formulated as:
where ψ represents the mother wavelet, t is time, a is a scale factor inversely proportional to frequency, and b is a shift factor.
It should be noted that the wavelet transform layer may extract bearing failure features from different angles using a plurality of wavelet bases. When the SWB works, bearing vibration signals are firstly input into the wavelet transformation layer, the input signals are subjected to wavelet transformation in the wavelet channel to decompose the bearing vibration signals, noise interference is reduced, and bearing fault characteristics are extracted.
The wavelet transform provides a time-frequency window which changes with the frequency of the signal, overcomes the defect that the size of the short-time Fourier transform window does not change with the frequency, and can be used for time-frequency analysis and processing of the signal. Therefore, the wavelet transform has the characteristic of multi-resolution analysis, and can perform localized analysis on the bearing vibration signal in the time domain and the frequency domain at the same time. The signal may be decomposed into subbands with multiple dimensions and frequencies. The signal has a high time resolution and a low frequency resolution at high frequencies and a high frequency resolution and a low time resolution at low frequencies, and the wavelet transform can characterize both time and frequency information. The multi-resolution nature of wavelet transform makes it excellent in terms of signal decomposition, denoising, and analysis of non-stationary signals.
Wavelets are a class of functions that are localized in both the time and frequency domains. Each wavelet is a variation of a single mother wavelet ψ, known as a wavelet member. Wavelet base { ψ } a,b (t) } is made up of a plurality of wavelet members, and can be constructed by controlling the scale factor a and the shifting factor b of the mother wavelet ψ (·). The scale factor a may cause the signal to contract or stretch. When the scale factor is low, the signal contracts. When the scale factor is high, the signal stretches. The wavelet transform comprises three steps:
(1) A wavelet basis function is selected and then coefficients of similarity to the partial signal are calculated.
(2) The wavelet is shifted to the right b and the calculation of the similarity coefficients is continued until all parts of the signal are calculated.
(3) Changing the scale parameter a of the wavelet, and repeating the steps (1) and (2) until all scale analysis is completed.
According to an embodiment of the present invention, the weighting and fusing the bearing fault characteristics of the multiple scales to obtain a fused fault characteristic includes:
multiplying the bearing fault characteristics of the multiple scales by corresponding influence weights respectively to obtain weight scores of the multiple fault characteristics;
selecting a plurality of key fault characteristics according to the weight scores of the plurality of fault characteristics;
and carrying out feature fusion on the plurality of key fault features to obtain fusion fault features.
It should be noted that the weighted fusion layer is a 1×1 standard convolution layer, and is used for weighted fusion of the characteristics of different wavelet channel outputs. Bearing failure characteristics from different wavelet channel outputs are of different importance due to the different wavelet basis functions in the wavelet channels. Then, the weighted fusion layer gives different weights to the output bearing fault characteristics through a group of 1×1 standard convolution layers, highlights key fault characteristics, and therefore determines the optimal bearing fault characteristics and fuses the characteristics. Finally, the SWB outputs a group of fault characteristics after weighted fusion, and takes the group of characteristics as the input of the CNN network, thereby enhancing the noise immunity of the CNN network.
Fig. 3 shows a flowchart of a bearing failure diagnosis information acquisition method of the present invention.
As shown in fig. 3, according to an embodiment of the present invention, the analyzing according to the fused fault characteristics to obtain bearing fault diagnosis information includes:
s302, inputting the fusion fault characteristics into a plurality of convolution layers;
s304, the plurality of convolution layers process the fusion fault characteristics according to a preset method and send the processed fault characteristics to a full connection layer;
And S306, the full-connection layer integrates and classifies the processed fault characteristics to obtain bearing fault diagnosis information.
Note that SWCNN includes one ultra-wavelet block, 4 standard convolution layers, some operations such as batch normalization, reLu and global max pooling, and one full connection layer. The preset method is specifically that the output of the ultra-wavelet block is used for enhancing the characteristic expression through a plurality of convolution layers. Then, operations such as batch normalization, reLu, and global max pooling are performed sequentially on the output of each layer of convolutional layers. And finally, outputting the classification result through the full connection layer.
In addition, the second-layer convolution of SWCNN adopts a wide convolution kernel, so that the receptive field is increased, and the diagnosis precision is improved. The subsequent convolution layers all adopt small convolution kernels, so that deep information hidden among samples can be deeply mined, and the number of network parameters is small. In addition, a batch normalization and pooling layer is added after all convolution layers to speed up training and reduce the number of parameters.
According to an embodiment of the present invention, further comprising:
the vibration signal sample data is expanded by means of overlap sampling.
It should be noted that, because of the multifunctional bearing fault simulation test stand, each fault data set has only 983040 sampling points, and the training samples are limited. When the training samples are too small, this may result in a model under-fit, and the accuracy of the training and test sets may be reduced. To avoid the problem of under-fitting, the scheme adopts an overlap sampling method to expand the data set. The sliding window size is 2048, and the sliding step size (offset) is 100. Starting from the starting position, sliding 100 sampling points backwards each time, the sliding window collects 2048 sampling points to sequentially construct data samples until 600 samples are available.
According to an embodiment of the present invention, further comprising:
in the training process of the preset ultra-wavelet convolutional neural network model, training data are processed through a cross entropy function and an Adam algorithm.
In order to improve the performance of the model, the experimental result reaches an optimal value, and the model parameters are selected according to the scheme. In the training strategy of the scheme, a cross entropy function and Adam are selected as a loss function and an optimizer so as to accelerate the convergence rate of the model and obtain an optimal result. Cross entropy is a loss function used to evaluate the difference between the predicted probability distribution and the actual distribution. The smaller the cross entropy, the closer the two probability distributions are, and the higher the accuracy of the model. Adam is a learning rate self-adaptive optimization algorithm, and by dynamically adjusting learning rate parameters, adam algorithm can improve the robustness of learning rate and help model convergence. Training the CNN model requires setting the learning rate within an appropriate range. Too high a learning rate may lead to instability, while too low a learning rate increases training time. Under the premise of guaranteeing training stability, training time is reduced, and the learning rate of the scheme is set to be 0.001. Meanwhile, the training batch size is 128.
Fig. 4 shows a block diagram of an ultra-wavelet based bearing failure diagnosis system of the present invention.
As shown in fig. 4, a second aspect of the present invention provides a bearing fault diagnosis system 4 based on ultra-wavelet, comprising a memory 41 and a processor 42, wherein the memory includes a bearing fault diagnosis method program based on ultra-wavelet, and the bearing fault diagnosis method program based on ultra-wavelet realizes the following steps when executed by the processor:
acquiring a bearing vibration signal;
inputting the bearing vibration signal into a preset ultra-wavelet convolutional neural network model, and analyzing the bearing vibration signal through an ultra-wavelet block to obtain bearing fault characteristics of multiple scales;
weighting and fusing the bearing fault characteristics of the multiple scales to obtain fused fault characteristics;
and analyzing according to the fusion fault characteristics to obtain bearing fault diagnosis information.
According to the embodiment of the invention, the preset ultra-wavelet convolutional neural network (SWCNN) is established according to the wavelet transformation principle. The SWCNN takes the ultra-wavelet convolution block as a first layer of the CNN network, so that the noise of the bearing vibration signal is reduced, and the key fault characteristics are enhanced. The scheme selects Morlet wavelet, laplace wavelet and Mexhat wavelet to form wavelet core of ultra-wavelet block. The ultra-wavelet block adaptively selects and constructs a wavelet basis which can effectively represent a bearing fault signal from an original bearing vibration signal, and simultaneously adopts a plurality of wavelet basis functions to carry out wavelet transformation, so that bearing fault characteristics are highlighted on different scales, and multi-scale fault characteristics are provided for a CNN model. In addition, the ultra-wavelet block can decompose key characteristic information from noise through the characteristic of wavelet transformation multi-resolution analysis, so that the noise immunity of the CNN model is improved.
SWCNN adopts Adaptive Wavelet Transform (AWT) and adopts Laplace wavelet, morlet wavelet and Mexh wavelet to form adaptive wavelet, so that the advantages of different wavelet bases can be fused. In addition, SWCNN does not need to manually select a wavelet base, and the self-adaptive wavelet is optimized by CNN on wavelet parameters and combination coefficients, so that a group of self-adaptive wavelet filters with good noise reduction performance are learned. The present protocol analyzes SWCNN through two different bearing datasets. The result shows that the SWCNN can well retain the original information, and fully considers the internal relation of the sequence data, thereby realizing fault diagnosis under the condition of strong noise.
According to an embodiment of the present invention, further comprising:
establishing an initial ultra-wavelet convolutional neural network model;
acquiring vibration signal sample data;
dividing the vibration signal sample data into a training set, a verification set and a test set according to a preset proportion;
and respectively inputting the training set, the verification set and the test set into the initial ultra-wavelet convolutional neural network model for training and diagnosis to obtain a preset ultra-wavelet convolutional neural network model.
The bearing data set of the experiment in this scheme is from a multifunctional bearing fault simulation test stand, and the experimental apparatus is shown in fig. 5. The experiment table mainly comprises a driving motor, a transmission part, a supporting part, a controller and the like. In the rolling bearing fault experiment, four signals are collected, and a normal bearing, an inner ring fault bearing, an outer ring fault bearing and a rolling body fault bearing are obtained. The sampling frequency is 8192Hz, the acquisition time is 120s, and the length of the sampling signal sequence of each fault condition is 983040.
Faults such as bearing cracks, pitting corrosion and the like in actual working conditions are simulated, faults with different depths are implanted into an inner ring, an outer ring and a rolling body of the rolling bearing in a mode of linear cutting and groove machining, and the depths are 0.4mm, 0.8mm and 1.2mm respectively. Healthy bearings can be considered a special failure mode, so there are ten types of bearing failure in the experimental dataset.
The training flow of the SWCNN model is shown in FIG. 6, and comprises 3 steps of data set division, training of the model and model diagnosis, wherein the specific steps are as follows:
(1) Dividing the original data, namely the vibration signal sample data, into three data sets according to a preset proportion: training set, verification set and test set, preset proportion is 8:1:1. wherein the data of the training set is raw noise-free data so that SWCNN can learn the appropriate wavelet basis. And the verification set and the test set are added with noise with a certain signal to noise ratio so as to judge whether the trained SWCNN model can resist noise interference and diagnose bearing faults.
(2) When training the model, firstly, initializing network parameters of SWCNN. And then inputting a training set to train parameters of each layer of SWCNN, so that the SWCNN can fully learn the characteristics of the original data. And finally, checking the trained SWCNN model by using a verification set containing different noises, and judging whether the target accuracy is reached. Repeating the steps for a plurality of times, and selecting an optimal model as a diagnosis model.
(3) In the model diagnosis, the optimal model in the step (2) is firstly taken as a diagnosis model. The test set is then input into a diagnostic model. Finally, the result of the diagnostic model is used as a judgment standard for evaluating SWCNN performance.
According to an embodiment of the present invention, after the bearing vibration signal is input to a preset ultra-wavelet convolutional neural network model, the method further includes:
acquiring wavelet database data;
analyzing the wavelet base data to obtain a plurality of self-adaptive wavelets;
the calculation method of the adaptive wavelet is expressed as the following formula:
wherein,,is the nth wavelet basis function, a n Is the scale factor of the wavelet basis function, b n Is a shift of the wavelet basis functionFactor, w n Is the combination coefficient of the ultra-wavelet, k represents the kth wavelet base function.
It should be noted that the wavelet transform can decompose the bearing vibration signal to reduce the noise and interference of the bearing vibration signal, and can obtain more information by utilizing the characteristics of multi-resolution analysis. To take advantage of the characteristics of wavelet transform multi-resolution analysis, the present solution proposes a Super Wavelet Block (SWB). SWB consists of two convolutional layers. The first convolution layer is a wavelet transform layer and is formed by a plurality of different wavelet channels in parallel, and each wavelet channel represents a wavelet base. The second convolution layer is a weighted fusion layer, which is a 1×1 standard convolution layer, used to perform weighted fusion on the characteristics of different wavelet channel outputs.
The SWB principle is that fault characteristics are extracted from bearing vibration signals by utilizing adaptive wavelets, the adaptive wavelets adopt a multi-wavelet-base complementary mechanism, complementary characteristics are obtained by utilizing different wavelet bases, different weights are given to the characteristics, and the bearing fault characteristics are fully extracted. Furthermore, the adaptive wavelet satisfies the wavelet's tolerability condition, i.e. it is regarded as a wavelet itself. Therefore, SWB can merge the advantages of different wavelet bases, obtain more fault characteristic information, and learn new wavelet bases.
The SWB has two main advantages, namely the SWB performs feature extraction through a plurality of wavelet basis functions so as to match different fault features in the vibration signal, obtain more useful information and further increase the noise immunity of the CNN model. The second advantage is that SWB can directly input original vibration signal, well keep original information, reduce the original vibration signal information loss that adopts denoising preliminary treatment to cause.
According to an embodiment of the present invention, the analyzing the bearing vibration signal by the ultra-wavelet block to obtain bearing fault characteristics with multiple dimensions includes:
the ultra-wavelet block carries out wavelet transformation on the bearing vibration signals through the plurality of self-adaptive wavelets respectively, so that the ultra-wavelet block carries out decomposition and noise reduction on the bearing vibration signals through wavelet transformation to obtain noise reduction vibration signals;
Extracting features of the noise reduction vibration signals through the plurality of adaptive wavelets to obtain bearing fault features with a plurality of scales;
the wavelet transform is formulated as:
where ψ represents the mother wavelet, t is time, a is a scale factor inversely proportional to frequency, and b is a shift factor.
It should be noted that the wavelet transform layer may extract bearing failure features from different angles using a plurality of wavelet bases. When the SWB works, bearing vibration signals are firstly input into the wavelet transformation layer, the input signals are subjected to wavelet transformation in the wavelet channel to decompose the bearing vibration signals, noise interference is reduced, and bearing fault characteristics are extracted.
The wavelet transform provides a time-frequency window which changes with the frequency of the signal, overcomes the defect that the size of the short-time Fourier transform window does not change with the frequency, and can be used for time-frequency analysis and processing of the signal. Therefore, the wavelet transform has the characteristic of multi-resolution analysis, and can perform localized analysis on the bearing vibration signal in the time domain and the frequency domain at the same time. The signal may be decomposed into subbands with multiple dimensions and frequencies. The signal has a high time resolution and a low frequency resolution at high frequencies and a high frequency resolution and a low time resolution at low frequencies, and the wavelet transform can characterize both time and frequency information. The multi-resolution nature of wavelet transform makes it excellent in terms of signal decomposition, denoising, and analysis of non-stationary signals.
Wavelets are a class of functions that are localized in both the time and frequency domains. Each wavelet is a variation of a single mother wavelet ψ, known as a wavelet member. Wavelet base { ψ } a,b (t) } is made up of a plurality of wavelet members, and can be constructed by controlling the scale factor a and the shifting factor b of the mother wavelet ψ (·). The scale factor a may cause the signal to contract or stretch. When the scale factor is low, the signal contracts. When the scale factor is high, the signal stretches. Wavelet transform comprisesThree steps:
(1) A wavelet basis function is selected and then coefficients of similarity to the partial signal are calculated.
(2) The wavelet is shifted to the right b and the calculation of the similarity coefficients is continued until all parts of the signal are calculated.
(3) Changing the scale parameter a of the wavelet, and repeating the steps (1) and (2) until all scale analysis is completed.
According to an embodiment of the present invention, the weighting and fusing the bearing fault characteristics of the multiple scales to obtain a fused fault characteristic includes:
multiplying the bearing fault characteristics of the multiple scales by corresponding influence weights respectively to obtain weight scores of the multiple fault characteristics;
selecting a plurality of key fault characteristics according to the weight scores of the plurality of fault characteristics;
and carrying out feature fusion on the plurality of key fault features to obtain fusion fault features.
It should be noted that the weighted fusion layer is a 1×1 standard convolution layer, and is used for weighted fusion of the characteristics of different wavelet channel outputs. Bearing failure characteristics from different wavelet channel outputs are of different importance due to the different wavelet basis functions in the wavelet channels. Then, the weighted fusion layer gives different weights to the output bearing fault characteristics through a group of 1×1 standard convolution layers, highlights key fault characteristics, and therefore determines the optimal bearing fault characteristics and fuses the characteristics. Finally, the SWB outputs a group of fault characteristics after weighted fusion, and takes the group of characteristics as the input of the CNN network, thereby enhancing the noise immunity of the CNN network.
According to an embodiment of the present invention, the analyzing according to the fused fault characteristics to obtain bearing fault diagnosis information includes:
inputting the fused fault signature to a plurality of convolution layers;
the plurality of convolution layers process the fusion fault characteristics according to a preset method and send the processed fault characteristics to a full connection layer;
and the full-connection layer integrates and classifies the processed fault characteristics to obtain bearing fault diagnosis information.
Note that SWCNN includes one ultra-wavelet block, 4 standard convolution layers, some operations such as batch normalization, reLu and global max pooling, and one full connection layer. The preset method is specifically that the output of the ultra-wavelet block is used for enhancing the characteristic expression through a plurality of convolution layers. Then, operations such as batch normalization, reLu, and global max pooling are performed sequentially on the output of each layer of convolutional layers. And finally, outputting the classification result through the full connection layer.
In addition, the second-layer convolution of SWCNN adopts a wide convolution kernel, so that the receptive field is increased, and the diagnosis precision is improved. The subsequent convolution layers all adopt small convolution kernels, so that deep information hidden among samples can be deeply mined, and the number of network parameters is small. In addition, a batch normalization and pooling layer is added after all convolution layers to speed up training and reduce the number of parameters.
According to an embodiment of the present invention, further comprising:
the vibration signal sample data is expanded by means of overlap sampling.
It should be noted that, because of the CWRU data sets, each failure data set has only 12 ten thousand sampling points, and the training samples are limited. When the training samples are too small, this may result in a model under-fit, and the accuracy of the training and test sets may be reduced. To avoid the problem of under-fitting, the scheme adopts an overlap sampling method to expand the data set. The sliding window size is 2048, and the sliding step size (offset) is 100. Starting from the starting position, sliding 100 sampling points backwards each time, the sliding window collects 2048 sampling points to sequentially construct data samples until 600 samples are available.
According to an embodiment of the present invention, further comprising:
in the training process of the preset ultra-wavelet convolutional neural network model, training data are processed through a cross entropy function and an Adam algorithm.
In order to improve the performance of the model, the experimental result reaches an optimal value, and the model parameters are selected according to the scheme. In the training strategy of the scheme, a cross entropy function and Adam are selected as a loss function and an optimizer so as to accelerate the convergence rate of the model and obtain an optimal result. Cross entropy is a loss function used to evaluate the difference between the predicted probability distribution and the actual distribution. The smaller the cross entropy, the closer the two probability distributions are, and the higher the accuracy of the model. Adam is a learning rate self-adaptive optimization algorithm, and by dynamically adjusting learning rate parameters, adam algorithm can improve the robustness of learning rate and help model convergence. Training the CNN model requires setting the learning rate within an appropriate range. Too high a learning rate may lead to instability, while too low a learning rate increases training time. Under the premise of guaranteeing training stability, training time is reduced, and the learning rate of the scheme is set to be 0.001. Meanwhile, the training batch size is 128.
A third aspect of the present invention provides a computer-readable storage medium having embodied therein a bearing failure diagnosis method program based on ultra-wavelets, which when executed by a processor, implements the steps of a bearing failure diagnosis method based on ultra-wavelets as described in any one of the above.
The invention discloses a fan gear box bearing fault diagnosis method, a system and a storage medium based on ultra-wavelet, wherein the method comprises the following steps: acquiring a bearing vibration signal; inputting the bearing vibration signal into a preset ultra-wavelet convolutional neural network model, and analyzing the bearing vibration signal through an ultra-wavelet block to obtain bearing fault characteristics of multiple scales; weighting and fusing the bearing fault characteristics of the multiple scales to obtain fused fault characteristics; and analyzing according to the fusion fault characteristics to obtain bearing fault diagnosis information. According to the invention, the bearing vibration signal is processed through the SWCNN, the bearing vibration signal containing noise is processed through the ultra-wavelet block, and the noise reduction processing and the feature extraction are respectively carried out by using 3 different wavelet basis functions, so that the noise resistance of the SWCNN network is enhanced. In addition, the weight fusion layer of the ultra-wavelet block can adaptively determine fusion weights, so that wavelet channels with obvious fault characteristics are enhanced, key characteristics are highlighted, and the diagnosis accuracy of SWCNN is improved.
In the several embodiments provided in this application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above described device embodiments are only illustrative, e.g. the division of the units is only one logical function division, and there may be other divisions in practice, such as: multiple units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. In addition, the various components shown or discussed may be coupled or directly coupled or communicatively coupled to each other via some interface, whether indirectly coupled or communicatively coupled to devices or units, whether electrically, mechanically, or otherwise.
The units described above as separate components may or may not be physically separate, and components shown as units may or may not be physical units; can be located in one place or distributed to a plurality of network units; some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in each embodiment of the present invention may be integrated in one processing unit, or each unit may be separately used as one unit, or two or more units may be integrated in one unit; the integrated units may be implemented in hardware or in hardware plus software functional units.
Those of ordinary skill in the art will appreciate that: all or part of the steps for implementing the above method embodiments may be implemented by hardware related to program instructions, and the foregoing program may be stored in a computer readable storage medium, where the program, when executed, performs steps including the above method embodiments; and the aforementioned storage medium includes: a mobile storage device, a Read-only memory (ROM), a random access memory (RAM, randomAccessMemory), a magnetic disk or an optical disk, or the like, which can store program codes.
Alternatively, the above-described integrated units of the present invention may be stored in a computer-readable storage medium if implemented in the form of software functional modules and sold or used as separate products. Based on such understanding, the technical solutions of the embodiments of the present invention may be embodied in essence or a part contributing to the prior art in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a removable storage device, ROM, RAM, magnetic or optical disk, or other medium capable of storing program code.
Claims (10)
1. A fan gear box bearing fault diagnosis method based on ultra-wavelet is characterized by comprising the following steps:
acquiring a bearing vibration signal;
inputting the bearing vibration signal into a preset ultra-wavelet convolutional neural network model, and analyzing the bearing vibration signal through an ultra-wavelet block to obtain bearing fault characteristics of multiple scales;
weighting and fusing the bearing fault characteristics of the multiple scales to obtain fused fault characteristics;
and analyzing according to the fusion fault characteristics to obtain bearing fault diagnosis information.
2. The method for diagnosing a bearing failure of a wind power motor based on ultra-wavelet as claimed in claim 1, further comprising:
establishing an initial ultra-wavelet convolutional neural network model;
acquiring fan gear box bearing vibration signal sample data;
dividing the vibration signal sample data into a training set, a verification set and a test set according to a preset proportion;
and respectively inputting the training set, the verification set and the test set into the initial ultra-wavelet convolutional neural network model for training and diagnosis to obtain a preset ultra-wavelet convolutional neural network model.
3. The method for diagnosing a bearing failure of a gearbox of a fan based on ultra-wavelet of claim 1, further comprising, after inputting the bearing vibration signal to a preset ultra-wavelet convolutional neural network model:
Acquiring a plurality of wavelet base data;
analyzing the wavelet base data to obtain a plurality of self-adaptive wavelets;
the calculation method of the adaptive wavelet is expressed as the following formula:
wherein,,is the nth wavelet basis function, a n Is the scale factor of the wavelet basis function, b n Is a panning factor of the wavelet basis function, w n Is the combination coefficient of the ultra-wavelet, k represents the kth wavelet base function.
4. The method for diagnosing bearing faults of a fan gearbox based on ultra-small waves according to claim 3, wherein the bearing vibration signals are analyzed through ultra-small wave blocks to obtain bearing fault characteristics with multiple scales, and the method comprises the following steps:
the ultra-wavelet block carries out wavelet transformation on the bearing vibration signals through the plurality of self-adaptive wavelet bases respectively, so that the ultra-wavelet block carries out decomposition and noise reduction on the bearing vibration signals through wavelet transformation to obtain noise reduction vibration signals;
extracting features of the noise reduction vibration signals through the plurality of adaptive wavelets to obtain bearing fault features with a plurality of scales;
the wavelet transform is formulated as:
where ψ represents the mother wavelet, t is time, a is a scale factor inversely proportional to frequency, and b is a shift factor.
5. The method for diagnosing a bearing failure of a gearbox of a fan based on ultra-wavelet according to claim 1, wherein the weighting and fusing the bearing failure characteristics of the multiple scales to obtain fused failure characteristics comprises:
multiplying the bearing fault characteristics of the multiple scales by corresponding influence weights respectively to obtain weight scores of the multiple fault characteristics;
selecting a plurality of key fault characteristics according to the weight scores of the plurality of fault characteristics;
and carrying out feature fusion on the plurality of key fault features to obtain fusion fault features.
6. The method for diagnosing bearing faults of a fan gearbox based on ultra-wavelet according to claim 1, wherein the analyzing according to the fused fault characteristics to obtain bearing fault diagnosis information comprises the following steps:
inputting the fused fault signature to a plurality of convolution layers;
the plurality of convolution layers process the fusion fault characteristics according to a preset method and send the processed fault characteristics to a full connection layer;
and the full-connection layer integrates and classifies the processed fault characteristics to obtain bearing fault diagnosis information.
7. The bearing fault diagnosis system based on the ultra-wavelet is characterized by comprising a memory and a processor, wherein the memory comprises a bearing fault diagnosis method program based on the ultra-wavelet, and the bearing fault diagnosis method program based on the ultra-wavelet realizes the following steps when being executed by the processor:
Acquiring a fan gear box bearing vibration signal;
inputting the bearing vibration signal into a preset ultra-wavelet convolutional neural network model, and analyzing the bearing vibration signal through an ultra-wavelet block to obtain bearing fault characteristics of multiple scales;
weighting and fusing the bearing fault characteristics of the multiple scales to obtain fused fault characteristics;
and analyzing according to the fusion fault characteristics to obtain bearing fault diagnosis information.
8. The system of claim 7, wherein after inputting the bearing vibration signal to a predetermined ultra-wavelet convolutional neural network model, further comprising:
acquiring a plurality of wavelet base data;
analyzing the wavelet base data to obtain a plurality of self-adaptive wavelets;
the calculation method of the adaptive wavelet is expressed as the following formula:
wherein,,is the nth wavelet basis function, a n Is the scale factor of the wavelet basis function, b n Is a panning factor of the wavelet basis function, w n Is the combination coefficient of the ultra-wavelet, k represents the kth wavelet base function.
9. The system for diagnosing a bearing failure of a gearbox of a wind turbine based on ultra-small waves of claim 8, wherein said analyzing said bearing vibration signal by ultra-small wave block to obtain bearing failure characteristics of multiple scales comprises:
The ultra-wavelet block carries out wavelet transformation on the bearing vibration signals through the plurality of self-adaptive wavelets respectively, so that the ultra-wavelet block carries out decomposition and noise reduction on the bearing vibration signals through wavelet transformation to obtain noise reduction vibration signals;
extracting features of the noise reduction vibration signals through the plurality of adaptive wavelets to obtain bearing fault features with a plurality of scales;
the wavelet transform is formulated as:
where ψ represents the mother wavelet, t is time, a is a scale factor inversely proportional to frequency, and b is a shift factor.
10. A computer-readable storage medium, characterized in that it includes therein a bearing failure diagnosis method program based on ultra-wavelets, which, when executed by a processor, implements the steps of a bearing failure diagnosis method based on ultra-wavelets as claimed in any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310462774.5A CN116484176A (en) | 2023-04-26 | 2023-04-26 | Bearing fault diagnosis method, system and storage medium based on ultra-wavelet |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310462774.5A CN116484176A (en) | 2023-04-26 | 2023-04-26 | Bearing fault diagnosis method, system and storage medium based on ultra-wavelet |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116484176A true CN116484176A (en) | 2023-07-25 |
Family
ID=87226554
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310462774.5A Pending CN116484176A (en) | 2023-04-26 | 2023-04-26 | Bearing fault diagnosis method, system and storage medium based on ultra-wavelet |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116484176A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117909668A (en) * | 2024-03-19 | 2024-04-19 | 安徽大学 | Bearing fault diagnosis method and system based on convolutional neural network |
-
2023
- 2023-04-26 CN CN202310462774.5A patent/CN116484176A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117909668A (en) * | 2024-03-19 | 2024-04-19 | 安徽大学 | Bearing fault diagnosis method and system based on convolutional neural network |
| CN117909668B (en) * | 2024-03-19 | 2024-06-07 | 安徽大学 | Bearing fault diagnosis method and system based on convolutional neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109946389B (en) | Structural damage identification method based on ensemble empirical mode decomposition and convolutional neural network | |
| Wang et al. | Fault diagnosis of a rolling bearing using wavelet packet denoising and random forests | |
| CN110516305B (en) | Intelligent fault diagnosis method under small sample based on attention mechanism meta-learning model | |
| CN111238814B (en) | Rolling bearing fault diagnosis method based on short-time Hilbert transform | |
| CN111126819B (en) | Intelligent analysis method for urban driving condition | |
| CN116226646B (en) | Method, system, equipment and medium for predicting health state and residual life of bearing | |
| Liu et al. | Multi-feature fusion for fault diagnosis of rotating machinery based on convolutional neural network | |
| CN111914705A (en) | Signal generation method and device for improving health state evaluation accuracy of reactor | |
| CN112926644A (en) | Method and system for predicting residual service life of rolling bearing | |
| CN111680788A (en) | Equipment fault diagnosis method based on deep learning | |
| CN116484176A (en) | Bearing fault diagnosis method, system and storage medium based on ultra-wavelet | |
| CN116304861A (en) | Time-frequency characteristic fusion fault diagnosis method based on self-attention | |
| CN113705424A (en) | Performance equipment fault diagnosis model construction method based on time convolution noise reduction network | |
| CN115290326A (en) | Rolling bearing fault intelligent diagnosis method | |
| CN113203914A (en) | Underground cable early fault detection and identification method based on DAE-CNN | |
| CN117030263A (en) | Bearing fault diagnosis method based on improved residual error network under multi-sensor signal fusion | |
| CN116702076A (en) | Small sample migration learning fault diagnosis method, system, computer and storage medium based on CNN feature fusion | |
| CN114459760B (en) | Rolling bearing fault diagnosis method and system in strong noise environment | |
| CN114065809A (en) | Method and device for identifying abnormal sound of passenger car, electronic equipment and storage medium | |
| CN114563671A (en) | High-voltage cable partial discharge diagnosis method based on CNN-LSTM-Attention neural network | |
| CN110222390B (en) | Gear crack identification method based on wavelet neural network | |
| CN110044619B (en) | Multi-fault feature identification method based on sparse multi-cycle group lasso | |
| CN116758922A (en) | Voiceprint monitoring and diagnosing method for transformer | |
| CN113409213B (en) | Method and system for enhancing noise reduction of time-frequency diagram of fault signal of plunger pump | |
| Li et al. | A robust fault diagnosis method for rolling bearings based on deep convolutional neural network |
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 |