CN116337449A - Sparse self-coding fault diagnosis method and system based on information fusion - Google Patents
Sparse self-coding fault diagnosis method and system based on information fusion Download PDFInfo
- Publication number
- CN116337449A CN116337449A CN202310294325.4A CN202310294325A CN116337449A CN 116337449 A CN116337449 A CN 116337449A CN 202310294325 A CN202310294325 A CN 202310294325A CN 116337449 A CN116337449 A CN 116337449A
- Authority
- CN
- China
- Prior art keywords
- fault diagnosis
- coding
- model
- sparse self
- 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 73
- 230000004927 fusion Effects 0.000 title claims abstract description 28
- 238000000034 method Methods 0.000 title claims abstract description 25
- 230000001133 acceleration Effects 0.000 claims abstract description 49
- 238000006073 displacement reaction Methods 0.000 claims abstract description 45
- 238000012549 training Methods 0.000 claims abstract description 44
- 238000001228 spectrum Methods 0.000 claims abstract description 37
- 238000012360 testing method Methods 0.000 claims abstract description 28
- 230000010354 integration Effects 0.000 claims description 9
- 238000005096 rolling process Methods 0.000 abstract description 8
- 230000006870 function Effects 0.000 description 15
- 239000002131 composite material Substances 0.000 description 9
- 230000004913 activation Effects 0.000 description 8
- 239000013598 vector Substances 0.000 description 7
- 230000008569 process Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000000605 extraction Methods 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000013136 deep learning model Methods 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000000007 visual 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
- 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
- G06F18/2136—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on sparsity criteria, e.g. with an overcomplete basis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- 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
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/251—Fusion techniques of input or preprocessed data
-
- 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
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- 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
- G06N3/047—Probabilistic or stochastic networks
-
- 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
- G06N3/048—Activation functions
-
- 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
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Acoustics & Sound (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a sparse self-coding fault diagnosis method and system based on information fusion, comprising the following steps: collecting vibration acceleration signals under different fault states; according to the acceleration signal, obtaining a frequency spectrum, a speed and a displacement signal; splicing and fusing the frequency spectrum, the speed and the displacement signals into a sample; dividing a data set constructed by a plurality of samples into a training set sample and a test set sample; inputting the training set with the label into a sparse self-coding model for fault diagnosis, and pre-training parameters of the sparse self-coding model for fault diagnosis; inputting a test set sample without a label into a stored pre-training model for testing; the saved model is used for fault diagnosis. The invention can realize fault diagnosis of the rolling bearing and the like, can reduce the network layer number and improve the fault diagnosis accuracy without manually extracting or defining the characteristics.
Description
Technical Field
The invention relates to a sparse self-coding fault diagnosis method based on information fusion, and belongs to the field of fault diagnosis.
Background
The rolling bearing fault diagnosis method based on deep learning is a research hotspot in the current fault diagnosis field, and the self-encoder is widely applied to the field of fault feature extraction due to strong feature extraction and reconstruction capability of the self-encoder on signals.
At present, most fault diagnosis models rely on a single vibration acceleration signal as an original input signal, the feature extraction of the single vibration acceleration signal has high requirement on priori knowledge, and the models can not fully utilize time domain and frequency domain information, so that the fault diagnosis effect of the deep learning model based on the single input acceleration signal is poor.
According to the invention, the frequency domain integration of the vibration acceleration signal is utilized to obtain the speed and displacement signal, the frequency spectrum of the combined acceleration signal is fused into a composite signal and is used as the input of a sparse self-coding network, the sparse self-coding fault diagnosis model based on information fusion is obtained through training, the number of network layers is reduced, the problem of insufficient utilization of time domain and frequency domain information is overcome, and therefore, the effective diagnosis of the rolling bearing is completed.
Disclosure of Invention
The invention provides a sparse self-coding fault diagnosis method and system based on information fusion, which are characterized in that a speed signal and a displacement signal are obtained by integrating frequency domains of vibration acceleration signals, the frequency spectrums of the combined acceleration signals are fused into a composite signal and serve as input of a sparse self-coding network, a sparse self-coding model for fault diagnosis based on information fusion is obtained through training, and the high accuracy of the fused signal in the process of participating in training the sparse self-coding model for fault diagnosis of rolling bearings and the like is further verified.
The technical scheme of the invention is as follows:
according to an aspect of the present invention, there is provided a sparse self-coding fault diagnosis method based on information fusion, including: collecting vibration acceleration signals under different fault states, and adding label information for the vibration acceleration signals according to the collected fault types; according to the acceleration signal, obtaining a frequency spectrum, a speed and a displacement signal; splicing and fusing the frequency spectrum, the speed and the displacement signals into a sample; dividing a data set constructed by a plurality of samples into a training set sample and a test set sample; inputting the training set with the label into a sparse self-coding model for fault diagnosis, and pre-training parameters of the sparse self-coding model for fault diagnosis; when the diagnosis accuracy reaches a first expected preset accuracy, finishing training and storing model parameters; inputting a test set sample without labels into a stored pre-training model for testing, outputting a prediction label set, comparing the prediction label set with a real label set, and storing the model if the second expected prediction accuracy is reached; the saved model is used for fault diagnosis.
The obtaining of the frequency spectrum, the speed and the displacement signals according to the acceleration signals comprises the following steps: performing primary and secondary frequency domain integration on the acceleration signal to obtain a speed and displacement signal respectively; and then carrying out Fourier transform on the acceleration signal to obtain a frequency spectrum signal.
When the frequency spectrum, the speed and the displacement signals are spliced and fused into one sample, the speed and the displacement signals account for 15% -50% of the frequency spectrum signal points.
The sparse self-coding model for fault diagnosis mainly comprises a coding layer, a decoding layer and a softMax classification layer; the input of the coding layer is the fused signal, the decoding layer is the reconstructed input signal, the input layer of the SoftMax is the output signal of the decoding layer, and the output layer of the SoftMax is the fault type.
According to another aspect of the present invention, there is provided a sparse self-coding fault diagnosis system based on information fusion, including: the acquisition module acquires vibration acceleration signals under different fault states and adds label information to the vibration acceleration signals according to the acquired fault types; the acquisition module is used for acquiring frequency spectrum, speed and displacement signals according to the acceleration signals; the fusion module is used for splicing and fusing the frequency spectrum, the speed and the displacement signals into a sample; dividing a data set constructed by a plurality of samples into a training set sample and a test set sample; the pre-training module inputs the training set with the label into a sparse self-coding model for fault diagnosis, and pre-trains parameters of the sparse self-coding model for fault diagnosis; when the diagnosis accuracy reaches a first expected preset accuracy, finishing training and storing model parameters; the diagnosis module is used for inputting a test set sample without a label into the stored pre-training model for testing, outputting a prediction label set, comparing the prediction label set with a real label set, and storing the model if the second expected prediction accuracy is reached; the saved model is used for fault diagnosis.
According to another aspect of the present invention, there is provided a processor for running a program, wherein the program, when running, performs any one of the above sparse self-coding fault diagnosis methods based on information fusion.
The beneficial effects of the invention are as follows: aiming at the problem that the deep learning model is insufficient in utilization of single time domain and frequency domain information of the rolling bearing, the sparse self-coding fault diagnosis method based on information fusion is provided.
Drawings
FIG. 1 is a flow chart of a sparse self-coding fault diagnosis method based on information fusion;
FIG. 2 is a schematic diagram of the structure of an automatic encoder;
FIG. 3 is a feature scatter plot of acceleration signals as input to a sparse self-encoding network extracting features;
FIG. 4 is a feature scatter plot of a spectral signal as input to a sparse self-encoding network extracting features;
FIG. 5 is a feature scatter plot of the information fused composite signal as input to extract features from a sparse self-encoding network; wherein the speed and displacement signals all occupy 15 percent;
FIG. 6 is a feature scatter plot of the information fused composite signal as input to extract features from a sparse self-encoding network; wherein the speed and displacement signals all occupy 50 percent.
Detailed Description
The invention will be further described with reference to the drawings and examples, but the invention is not limited to the scope.
Example 1:
1-6, according to an aspect of the embodiment of the present invention, there is provided a sparse self-coding fault diagnosis method based on information fusion, including: collecting vibration acceleration signals of the rolling bearing under different fault states by adopting an acceleration sensor, and adding label information for the vibration acceleration signals according to the collected fault types; according to the acceleration signal, obtaining a frequency spectrum, a speed and a displacement signal; splicing and fusing the frequency spectrum, the speed and the displacement signals into a sample; dividing a data set constructed by a plurality of samples into a training set sample and a test set sample, adding label information into the training set sample, and independently labeling the label information of the test set sample; inputting the training set with the label into a sparse self-coding model for fault diagnosis, and pre-training parameters of the sparse self-coding model for fault diagnosis; when the diagnosis accuracy reaches a first expected preset accuracy, finishing training and storing model parameters; inputting a test set sample without labels into a stored pre-training model for testing, outputting a prediction label set, comparing the prediction label set with a real label set, and storing the model if the second expected prediction accuracy is reached; the saved model is used for fault diagnosis. In the embodiment of the invention, the first expected preset accuracy is set to be 100%, and the second expected preset accuracy is set to be 95%.
Further, the obtaining the frequency spectrum, the speed and the displacement signals according to the acceleration signals includes: performing primary and secondary frequency domain integration on the acceleration signal to obtain a speed and displacement signal respectively; and then carrying out Fourier transform on the acceleration signal to obtain a frequency spectrum signal.
Further, when the spectrum, speed and displacement signals are spliced and fused into one sample, the speed and displacement signals account for 15% -50% of the number of points of the spectrum signals.
Further, the sparse self-coding model for fault diagnosis mainly comprises a coding layer, a decoding layer and a softMax classification layer; the input of the coding layer is the fused signal, the decoding layer is the reconstructed input signal, the input layer of the SoftMax is the output signal of the decoding layer, and the output layer of the SoftMax is the fault type.
Further, in connection with experimental data, the present invention gives alternative embodiments as follows:
step (1): bearing vibration signal data disclosed by Kassi university of America are selected for testing, the vibration signals comprise bearing normal signals, bearing inner ring fault signals, bearing rolling body fault signals and bearing outer ring fault signals, the damage diameter of each fault state is 3 damage sizes of 0.1778, 0.3556 and 0.5334mm, so that 9 fault states and 1 normal state are divided into 10 states, and the 10 states are labeled. The vibration acceleration signal is converted into a frequency spectrum signal by using fast Fourier transform, and the data length is taken as 1024. And carrying out primary and secondary frequency domain integration on the acceleration signal to obtain a speed and displacement signal. In order to reduce the model training time and the effectiveness of discussing speed and displacement signal fusion on the model, only the length which occupies a certain proportion of the frequency spectrum information is selected for the speed and displacement signals after frequency domain integration to perform data fusion, namely the number of points of the selected speed and displacement signals occupies 15%, 30% and 50% of the number of points of the frequency spectrum signals, namely 3 modes of 154, 308 and 512 points are selected for each sample of the speed and displacement signals. After the signal segmentation processing, according to 7:3 to divide the training set and the test set.
Specifically, the fourier variable form of the acceleration signal a (t) at a certain frequency is expressed as:
a(t)=Ae jwt (1)
wherein a (t) is an acceleration signal, A is a coefficient corresponding to an acceleration signal component a (t), w represents frequency, t represents time, e jwt Representing a complex function.
When the initial speed is 0, the acceleration signal is time integrated to obtain a speed signal component, namely:
where V is a coefficient corresponding to the velocity signal component V (t).
The relationship of the primary integral in the frequency domain is:
when the initial velocity and the initial displacement component are both 0, the displacement signal component can be obtained by twice integrating the Fourier variable of the acceleration signal, namely:
where s (t) is a displacement signal, and X is a coefficient corresponding to the displacement signal component s (t).
The relationship of the twice integral in the frequency domain is:
and calculating Fourier components of all different frequencies in the vibration acceleration signal according to a relational expression integrated in a frequency domain, and then performing inverse Fourier transformation to obtain a corresponding speed signal and a displacement signal. The calculation result of the frequency domain integration is more accurate than that of the time domain integration algorithm, so that the frequency domain integration algorithm is selected to calculate the speed and displacement signals.
The fourier transform of the acceleration signal into a spectral signal is formulated as:
wherein F (t) is the frequency spectrum signal after Fourier transform.
And finally, merging the frequency spectrum signal, the speed signal and the displacement signal of the acceleration signal into a composite signal.
In this embodiment, the adopted data includes four states of a normal bearing, an inner ring fault, a rolling body fault and an outer ring fault, the load condition of the data samples of each state is the same, the sampling frequency is 12kHz, and the data selects the signal data of the driving end (DE end) of the motor. The data set sample length is 3 types, namely 1332, 1640 and 2048, each type of data set comprises 100 samples, and the training sample set and the test sample set are divided according to the proportion of 70% and 30%. The bearing sample set data table according to the description above is shown in table 1.
Table 1 bearing data sample set partitioning
(2) And a sparse self-coding model for fault diagnosis is built by adopting Matlab, and the model mainly comprises a coding layer, a decoding layer and a softMax classification layer. The input of the coding layer is the composite signal after fusion, the decoding layer is the reconstructed input signal, the input layer of the SoftMax is the output signal of the decoding layer, and the output layer of the SoftMax is the class number of faults. The specific process is as follows:
the self-encoder is an unsupervised self-adaptive learning neural network, and consists of an input layer, an hidden layer and an output layer, wherein hidden layer characteristics of input data are extracted through two processes of encoding and decoding of input, the sparse self-encoder adds coefficient penalty items on the basis of the change of an objective function of the self-encoder, and sparse data characteristics are extracted through the sparse penalty items, so that the dimension is effectively reduced and the clustering effect is improved.
Encoding process of the self-encoder: let the input unlabeled bearing sample data be: { x 1 ,x 2 ,…,x n-1 ,x n Using encoder activation function f θ Coding input data into hidden layer vector h m The method comprises the following steps:
wherein: f (f) θ The activation function of (2) is a tanh function; z represents an argument of an input activation function, which refers herein to input sample data; h is a m The vector representing the hidden layer after the input passes through the coding layer, the weight matrix and bias parameters of the coding layer are w and b.
The decoding process comprises the following steps: by decoding function g θ For hidden layer vector h m Reconstructing to obtain output vectorNamely:
wherein:is decoded by a decoding layerOutput vector g θ Is a decoding layer function, f θ' Is an activation function of the decoding layer; the weight matrix and bias parameters of the decoding layer are w' and c, respectively.
In the process of reconstructing vectors of an hidden layer, calculating errors of a network by constructing an error loss function, so as to obtain an optimal output result, wherein the loss function J is as follows:
wherein J is a loss function, x i ,The i-th sample of the input sample set and the i-th output vector of the decoding layer, respectively, and n is the number of samples.
Introducing a sparse penalty term into a loss function of a self-encoder, controlling the activation quantity of neurons of an hidden layer, selecting tanh as an activation function, and enabling the average activation quantity of the jth neurons of the hidden layerCan be expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,representing the average activation of the jth neuron of the hidden layer. The neuron outputs are active when they are close to 1 and inhibited when they are close to-1.
The difference between the 2 distributions is measured by the relative entropy, a relative sparsity factor ρ is introduced, and then a penalty factor can be defined as:
wherein KL (ρ||ρ) j ) For KL divergence, ρ is the sparsity factor.
Sparsity constraint is added to the self-encoder, and equation (12) is added as a penalty factor to the original loss function J.
Step (3): the composite signals with different lengths are respectively sent into a self-encoder network, and the lengths of the input data are respectively as follows: 1332. 1640 and 2048, setting the output layer of softmax as 10, setting the number of hidden layers input in three modes, adjusting the parameters of the network, and outputting the confusion matrix diagram for obtaining the training diagnosis accuracy.
Step (4): when the diagnosis accuracy of the training sample set confusion matrix is 100%, pre-storing parameters of the network model, inputting the test sample set into the stored model to obtain a prediction label set of the test sample, and analyzing whether the test accuracy reaches an expected target after comparing the prediction label set with a real label value. After the expected target is reached, repeating 10 times of experiments on the newly acquired data set containing 100 samples in each class, and averaging the obtained diagnosis accuracy to obtain the final diagnosis accuracy. The results obtained from 10 experiments are shown in Table 2.
Table 2 diagnostic results of three different ratio mixed composite signal inputs
Ratio of velocity and displacement signals | Average diagnostic accuracy | Variance of |
15% | 98.33% | 23.02 |
30% | 98.87% | 6.70 |
50% | 99.20% | 4.84 |
In order to compare the model diagnosis effects before and after adding the speed and displacement information, the spectrum signals are input into a 3-layer stacked sparse self-coding network model according to the network model training method, and the diagnosis accuracy obtained by 10 experiments is only 91.40%. Compared with the final result, after 15% of speed and displacement information is blended in the frequency spectrum signal, only one sparse self-coding layer is required to have the average accuracy of 98.33%, and the fault diagnosis of the bearing can be effectively carried out. In addition, the result under a certain experiment is randomly selected for visual display, as shown in fig. 3-6, it can be obviously seen from the graph that the acceleration signal is taken as input, the aliasing of the scatter diagram of the depth characteristic extracted by the sparse self-coding network is obvious, and 10 states of the bearing are completely inseparable; the frequency spectrum is used as input, the depth frequency spectrum characteristics extracted by the sparse self-coding network reduce the overlapping degree of different states, the characteristic scatter diagram of 10 states is basically separable, but has extremely small partial division errors, the distinction between the bearing states of different states is not obvious enough, the intra-class distance of the same state is large, and the inter-class distance of different states is small; the composite signals mixed by three different proportions are taken as input, 15% and 50% are taken as examples, the characteristic scatter diagrams of 10 states are obviously separable, the class inner distance of the same state is obviously reduced, and the class distance of different states is obviously increased.
The strong self-adaptive feature extraction capability of the self-encoder provides a thought for fault feature extraction, and the sparse self-encoding fault diagnosis model based on information fusion can not only fully utilize the information of the time domain and the frequency domain, but also adaptively extract the fault information features of the bearing, thereby reducing network complexity and effectively improving the diagnosis accuracy of the bearing.
According to another aspect of the embodiment of the present invention, there is provided a sparse self-coding fault diagnosis system based on information fusion, including: the acquisition module acquires vibration acceleration signals under different fault states and adds label information to the vibration acceleration signals according to the acquired fault types; the acquisition module is used for acquiring frequency spectrum, speed and displacement signals according to the acceleration signals; the fusion module is used for splicing and fusing the frequency spectrum, the speed and the displacement signals into a sample; dividing a data set constructed by a plurality of samples into a training set sample and a test set sample; the pre-training module inputs the training set with the label into a sparse self-coding model for fault diagnosis, and pre-trains parameters of the sparse self-coding model for fault diagnosis; when the diagnosis accuracy reaches a first expected preset accuracy, finishing training and storing model parameters; the diagnosis module is used for inputting a test set sample without a label into the stored pre-training model for testing, outputting a prediction label set, comparing the prediction label set with a real label set, and storing the model if the second expected prediction accuracy is reached; the saved model is used for fault diagnosis.
For portions of the foregoing that are not detailed for each module, reference may be made to the relevant description of the embodiments.
According to another aspect of the embodiment of the present invention, there is provided a processor, configured to execute a program, where the program executes any one of the above sparse self-coding fault diagnosis methods based on information fusion.
While the present invention has been described in detail with reference to the drawings, the present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention within the knowledge of those skilled in the art.
Claims (6)
1. The sparse self-coding fault diagnosis method based on information fusion is characterized by comprising the following steps of:
collecting vibration acceleration signals under different fault states, and adding label information for the vibration acceleration signals according to the collected fault types;
according to the acceleration signal, obtaining a frequency spectrum, a speed and a displacement signal;
splicing and fusing the frequency spectrum, the speed and the displacement signals into a sample; dividing a data set constructed by a plurality of samples into a training set sample and a test set sample;
inputting the training set with the label into a sparse self-coding model for fault diagnosis, and pre-training parameters of the sparse self-coding model for fault diagnosis; when the diagnosis accuracy reaches a first expected preset accuracy, finishing training and storing model parameters;
inputting a test set sample without labels into a stored pre-training model for testing, outputting a prediction label set, comparing the prediction label set with a real label set, and storing the model if the second expected prediction accuracy is reached; the saved model is used for fault diagnosis.
2. The sparse self-coding fault diagnosis method based on information fusion according to claim 1, wherein the obtaining a frequency spectrum, a velocity and a displacement signal according to an acceleration signal comprises: performing primary and secondary frequency domain integration on the acceleration signal to obtain a speed and displacement signal respectively; and then carrying out Fourier transform on the acceleration signal to obtain a frequency spectrum signal.
3. The sparse self-coding fault diagnosis method based on information fusion according to claim 1, wherein when the spectrum, speed and displacement signals are spliced and fused into one sample, the speed and displacement signals account for 15% -50% of the number of spectrum signals.
4. The sparse self-coding fault diagnosis method based on information fusion according to claim 1, wherein the sparse self-coding model for fault diagnosis mainly comprises a coding layer, a decoding layer and a SoftMax classification layer; the input of the coding layer is the fused signal, the decoding layer is the reconstructed input signal, the input layer of the SoftMax is the output signal of the decoding layer, and the output layer of the SoftMax is the fault type.
5. A sparse self-encoding fault diagnosis system based on information fusion, comprising:
the acquisition module acquires vibration acceleration signals under different fault states and adds label information to the vibration acceleration signals according to the acquired fault types;
the acquisition module is used for acquiring frequency spectrum, speed and displacement signals according to the acceleration signals;
the fusion module is used for splicing and fusing the frequency spectrum, the speed and the displacement signals into a sample; dividing a data set constructed by a plurality of samples into a training set sample and a test set sample;
the pre-training module inputs the training set with the label into a sparse self-coding model for fault diagnosis, and pre-trains parameters of the sparse self-coding model for fault diagnosis; when the diagnosis accuracy reaches a first expected preset accuracy, finishing training and storing model parameters;
the diagnosis module is used for inputting a test set sample without a label into the stored pre-training model for testing, outputting a prediction label set, comparing the prediction label set with a real label set, and storing the model if the second expected prediction accuracy is reached; the saved model is used for fault diagnosis.
6. A processor, wherein the processor is configured to run a program, and wherein the program, when run, performs the sparse self-encoding fault diagnosis method based on information fusion of any one of claims 1-4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310294325.4A CN116337449A (en) | 2023-03-24 | 2023-03-24 | Sparse self-coding fault diagnosis method and system based on information fusion |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310294325.4A CN116337449A (en) | 2023-03-24 | 2023-03-24 | Sparse self-coding fault diagnosis method and system based on information fusion |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116337449A true CN116337449A (en) | 2023-06-27 |
Family
ID=86887308
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310294325.4A Pending CN116337449A (en) | 2023-03-24 | 2023-03-24 | Sparse self-coding fault diagnosis method and system based on information fusion |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116337449A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116558828A (en) * | 2023-07-10 | 2023-08-08 | 昆明理工大学 | Rolling bearing health state assessment method based on autocorrelation coefficient sparsity characteristic |
CN117556344A (en) * | 2024-01-08 | 2024-02-13 | 浙江大学 | Fault diagnosis method and system for ball mill transmission system based on multi-source information fusion |
-
2023
- 2023-03-24 CN CN202310294325.4A patent/CN116337449A/en active Pending
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116558828A (en) * | 2023-07-10 | 2023-08-08 | 昆明理工大学 | Rolling bearing health state assessment method based on autocorrelation coefficient sparsity characteristic |
CN116558828B (en) * | 2023-07-10 | 2023-09-15 | 昆明理工大学 | Rolling bearing health state assessment method based on autocorrelation coefficient sparsity characteristic |
CN117556344A (en) * | 2024-01-08 | 2024-02-13 | 浙江大学 | Fault diagnosis method and system for ball mill transmission system based on multi-source information fusion |
CN117556344B (en) * | 2024-01-08 | 2024-05-14 | 浙江大学 | Fault diagnosis method and system for ball mill transmission system based on multi-source information fusion |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhao et al. | Improved generative adversarial network for vibration-based fault diagnosis with imbalanced data | |
CN116337449A (en) | Sparse self-coding fault diagnosis method and system based on information fusion | |
CN110057584B (en) | Degradation monitoring method for locomotive traction motor bearing | |
CN103728551B (en) | A kind of analog-circuit fault diagnosis method based on cascade integrated classifier | |
CN104714925B (en) | A kind of gear transmission noises analysis method based on Fourier Transform of Fractional Order and SVMs | |
CN111444988B (en) | Rolling bearing fault diagnosis system | |
CN113869208B (en) | Rolling bearing fault diagnosis method based on SA-ACWGAN-GP | |
CN111562108A (en) | Rolling bearing intelligent fault diagnosis method based on CNN and FCMC | |
Lu et al. | A novel feature extraction method using deep neural network for rolling bearing fault diagnosis | |
CN110243590B (en) | Rotor system fault diagnosis method based on principal component analysis and width learning | |
Liu et al. | Multi-feature fusion for fault diagnosis of rotating machinery based on convolutional neural network | |
CN114564982A (en) | Automatic identification method for radar signal modulation type | |
CN112541511A (en) | Multi-channel time series data fault diagnosis method based on convolutional neural network | |
CN114169377A (en) | G-MSCNN-based fault diagnosis method for rolling bearing in noisy environment | |
CN111507047A (en) | Inverse scattering imaging method based on SP-CUnet | |
CN114386452B (en) | Nuclear power circulating water pump sun gear fault detection method | |
Yang et al. | Few-shot learning for rolling bearing fault diagnosis via siamese two-dimensional convolutional neural network | |
Yi et al. | A bearing fault diagnosis method based on an improved depth residual network | |
Zuluaga et al. | Blind microscopy image denoising with a deep residual and multiscale encoder/decoder network | |
Singh et al. | Deep hidden analysis: A statistical framework to prune feature maps | |
CN113256443B (en) | Nuclear power water pump guide bearing fault detection method, system, equipment and readable storage medium | |
CN114021603A (en) | Radar signal modulation mode identification method based on model reparameterization | |
CN117221816A (en) | Multi-building floor positioning method based on Wavelet-CNN | |
CN113409213B (en) | Method and system for enhancing noise reduction of time-frequency diagram of fault signal of plunger pump | |
CN115166514A (en) | Motor fault identification method and system based on self-adaptive spectrum segmentation and denoising |
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 |