CN115374829A - Deep learning-based bearing fault diagnosis method and system - Google Patents
Deep learning-based bearing fault diagnosis method and system Download PDFInfo
- Publication number
- CN115374829A CN115374829A CN202211147956.5A CN202211147956A CN115374829A CN 115374829 A CN115374829 A CN 115374829A CN 202211147956 A CN202211147956 A CN 202211147956A CN 115374829 A CN115374829 A CN 115374829A
- Authority
- CN
- China
- Prior art keywords
- signal
- vibration acceleration
- equipment
- signals
- deep learning
- 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
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
-
- 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
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention provides a bearing fault diagnosis method based on deep learning, which comprises the following steps: acquiring vibration acceleration of X, Y and Z axes of industrial field mechanical equipment and equipment surface temperature data by a portable vibration signal acquisition device for 5G edge calculation; transmitting the vibration acceleration of X, Y and Z three axes of industrial field mechanical equipment and the surface temperature data of the equipment to a cloud end; after receiving the vibration acceleration and the surface temperature data, the cloud converts the three-axis vibration acceleration from a time domain to a frequency domain through Fourier transform; inputting the converted signals into a deep learning network for detection, wherein the number of input signal channels is four; taking the deep features extracted by the resnet18 network as input, and acquiring a learning dependence problem in signal sequence data through ON-LSTM; the time domain characteristics extracted by the ON-LSTM network are used as input, and the SVM receives the input signals, trains and outputs information expressed by the signals, normal signals and various fault signals to obtain specific fault types of the signals.
Description
Technical Field
The invention relates to the technical field of fault diagnosis, in particular to a bearing fault diagnosis method and system based on deep learning.
Background
The bearing is an indispensable key part in the rotating machinery, about 30% -40% of equipment faults are caused by the bearing faults in the fault detection of the equipment, the intelligent fault diagnosis of the motor bearing is always a research hotspot, and the early detection of newly-occurred faults is very important for the current complex system, so that the time and the cost can be saved, and necessary measures can be taken to avoid dangerous situations.
The conventional bearing fault diagnosis method can be divided into model-based, signal-based and expert knowledge-based methods, historical data is needed to establish a system fault model without a priori known model or signal mode, and an accurate model is difficult to establish or accurate effective signal characteristics are difficult to extract.
With the improvement of the intelligent manufacturing level, the state monitoring of electromechanical equipment advances into the 'big data' era, a large amount of production data is fully utilized, and a new opportunity is provided for a data-driven fault diagnosis method. The data hides abundant mechanical equipment operation state information. The deep learning method can directly extract the data internal relation and the complex relation between the fault characteristic quantity and the actual fault from a large amount of complex data. The end-to-end data fault diagnosis method is efficient and accurate.
Disclosure of Invention
Technical problem to be solved
Aiming at the defects of the prior art, the invention provides a bearing fault diagnosis method and equipment based on deep learning, and solves the problems that the conventional bearing fault diagnosis method proposed in the background art can be divided into model-based, signal-based and expert knowledge-based methods, and the methods need historical data to establish a system fault model without a priori known model or signal mode and are difficult to establish an accurate model or extract accurate and effective signal characteristics.
(II) technical scheme
In order to achieve the purpose, the invention is realized by the following technical scheme:
a bearing fault diagnosis method based on deep learning comprises the following steps:
acquiring vibration acceleration of X, Y and Z axes of industrial field mechanical equipment and equipment surface temperature data by a portable vibration signal acquisition device for 5G edge calculation;
transmitting the vibration acceleration of X, Y and Z axes of the industrial field mechanical equipment and the surface temperature data of the equipment to a cloud end;
after receiving the vibration acceleration and the surface temperature data, the cloud converts the three-axis vibration acceleration from a time domain to a frequency domain through Fourier transform;
inputting the converted signals into a deep learning network for detection, wherein the number of input signal channels is four;
taking the deep features extracted by the resnet18 network as input, and acquiring a learning dependence problem in signal sequence data through ON-LSTM;
the time domain characteristics extracted by the ON-LSTM network are used as input, and the support vector machine SVM receives an input signal, trains and then outputs information expressed by the signal, a normal signal and various fault signals to obtain the specific fault type of the signal.
Preferably, the portable vibration signal collector for 5G edge calculation is used for collecting vibration acceleration and equipment surface temperature data of X, Y and Z three axes of the industrial field mechanical equipment so as to collect equipment bearings in operation.
Preferably, the converted signal is input into a deep learning network for detection, and the number of input signal channels is four, which are the vibration acceleration of three axes X, Y, and Z and the temperature signal of the surface of the device.
The invention also provides a bearing fault diagnosis system based on deep learning, which comprises:
the signal acquisition module: the portable vibration signal collector is used for collecting vibration acceleration of X, Y and Z three axes of industrial field mechanical equipment and equipment surface temperature data through 5G edge calculation;
a data transmission module: the system is used for transmitting the vibration acceleration of the X, Y and Z three axes of the industrial field mechanical equipment and the surface temperature data of the equipment to a cloud end;
a deep learning module: the cloud end is used for receiving the vibration acceleration and the surface temperature data and then converting the three-axis vibration acceleration from a time domain to a frequency domain through Fourier transform;
inputting the converted signals into a deep learning network for detection, wherein the number of input signal channels is four;
taking the deep features extracted by the resnet18 network as input, and acquiring a learning dependence problem in signal sequence data through ON-LSTM;
a fault type output module: the method is used for taking the time domain characteristics extracted by the ON-LSTM network as input, and outputting information expressed by the signal, normal signals and various fault signals after the support vector machine SVM receives the input signal for training to obtain the specific fault type of the signal.
The present invention also provides a computer-readable storage medium, comprising: a stored program, wherein the program when executed performs the method as set forth in any of the preceding claims.
The invention also provides an electronic device comprising a memory having a computer program stored therein and a processor arranged to perform the method according to any of the preceding claims by means of the computer program.
(III) advantageous effects
The invention provides a bearing fault diagnosis method and system based on deep learning. The method has the following beneficial effects:
1. the signals collected by the bearing in the operation process can be collected in real time, and the fault diagnosis can be carried out in the operation process of the equipment, so that the operation efficiency of the equipment is improved.
2. Data are acquired from the cloud, so that a computer is prevented from acquiring equipment through a transmission signal line, and the site is cleaner and tidier.
3. The signals are converted from time domain to frequency domain through Fourier transform, so that the signal characteristics are more obvious, the characteristics are better separated, and the diagnosis accuracy is improved.
4. The input signal multi-channel information is beneficial to fault diagnosis by combining information of each channel, and compared with a single channel, the multi-channel information diagnosis is higher in accuracy and more stable.
5. The classification of the last fault by the support vector machine SVM is more accurate than softmax.
Drawings
FIG. 1 is a flow chart of a bearing fault diagnosis method based on deep learning according to the present invention;
fig. 2 is a structure of a bearing fault diagnosis device based on deep learning provided by the invention.
Detailed Description
The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.
As shown in fig. 1, a bearing fault diagnosis method based on deep learning according to an embodiment of the present invention includes:
s1, acquiring vibration acceleration of three axes X, Y and Z of industrial field mechanical equipment and equipment surface temperature data through a 5G edge calculation portable vibration signal acquisition device;
s2, transmitting the vibration acceleration of X, Y and Z axes of the industrial field mechanical equipment and the surface temperature data of the equipment to a cloud end;
s3, after receiving the vibration acceleration and the surface temperature data, the cloud converts the three-axis vibration acceleration from a time domain to a frequency domain through Fourier transform; the time domain is converted into the frequency domain, so that the fault characteristics are more obvious, and the diagnosis of the fault is facilitated;
in one embodiment, the fault signal may be characterized more prominently by converting the time domain signal to a frequency domain signal by fast fourier transforming the time domain signal.
S4, inputting the converted signals into a deep learning network for detection, wherein the number of input signal channels is four; compared with single-channel data, the data of three channels are added, and the precision of fault diagnosis is improved through complementation among the data of each channel;
s5, taking the deep features extracted by the resnet18 network as input, and acquiring a learning dependence problem in signal sequence data through ON-LSTM;
compared with the traditional CNN, the method introduces a residual block by taking the deep features extracted by the resnet18 network as input, solves the problem of network degradation of the deep network, and can construct a deeper network and extract deeper information.
ON-LSTM, wherein "ON" is collectively called "OrderedNeurons", i.e., ordered neurons, in other words neurons within such LSTM are specifically ordered, thereby allowing for the expression of richer information.
In contrast to LSTM, neurons inside ON-LSTM are specifically ordered to integrate a hierarchical structure (tree structure) into LSTM, allowing LSTM to automatically learn the hierarchical structure information, thereby enabling richer information to be expressed;
in one embodiment, the deep information extracted by Resnet18 is input into the ON-LSTM network to obtain the timing information corresponding to the signal.
In one embodiment, resnet18, linked by a residual, allows the network to be deeper, thereby extracting deeper levels of information from the vibration signal for ease of diagnosis.
S6, the time domain features extracted by the ON-LSTM network are used as input, the support vector machine SVM receives the input signals, trains and then outputs information expressed by the signals, normal signals and various fault signals, and specific fault types of the signals are obtained.
Compared with the traditional softmax, the SVM classifies the faults, and the problems that the model generalization capability is insufficient and the fault classification cannot be well adapted due to the fact that the softmax is used as a convolutional neural network classifier are solved;
in one embodiment, the SVM can convert the features into higher dimensions for classification, and the classification is more accurate.
In one embodiment, the signals are classified, the output of the ON-LSTM network is used as training data of the SVM network, parameters of the SVM network are set to start training, and specific faults of the signals are obtained.
Preferably, the portable vibration signal collector for calculating the 5G edge is used for collecting the vibration acceleration and the surface temperature data of the X, Y and Z three axes of the industrial field mechanical equipment so as to collect the equipment bearing in operation.
Preferably, the converted signal is input into a deep learning network for detection, and the number of input signal channels is four, which are the vibration acceleration of three axes X, Y, and Z and the temperature signal of the surface of the device.
As shown in fig. 2, the present invention further provides a deep learning-based bearing fault diagnosis system, including:
the signal acquisition module: the portable vibration signal collector is used for collecting vibration acceleration of X, Y and Z three axes of industrial field mechanical equipment and equipment surface temperature data through a 5G edge calculation portable vibration signal collector;
a data transmission module: the system is used for transmitting the vibration acceleration of the X, Y and Z three axes of the industrial field mechanical equipment and the surface temperature data of the equipment to a cloud end;
a deep learning module: the cloud end is used for receiving the vibration acceleration and the surface temperature data and then converting the three-axis vibration acceleration from a time domain to a frequency domain through Fourier transform;
inputting the converted signals into a deep learning network for detection, wherein the number of input signal channels is four;
taking the deep features extracted by the resnet18 network as input, and acquiring a learning dependence problem in signal sequence data through ON-LSTM;
a fault type output module: the method is used for taking the time domain characteristics extracted by the ON-LSTM network as input, and outputting information expressed by the signal, normal signals and various fault signals after the support vector machine SVM receives the input signal for training to obtain the specific fault type of the signal.
The present invention also provides a computer-readable storage medium, comprising: a stored program, wherein the program when executed performs the method of any preceding claim.
The invention also provides an electronic device comprising a memory having a computer program stored therein and a processor arranged to perform the method according to any of the preceding claims by means of the computer program.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (8)
1. A bearing fault diagnosis method based on deep learning is characterized by comprising the following steps:
acquiring vibration acceleration of X, Y and Z axes of industrial field mechanical equipment and equipment surface temperature data by a portable vibration signal acquisition device for 5G edge calculation;
transmitting the vibration acceleration of X, Y and Z axes of the industrial field mechanical equipment and the surface temperature data of the equipment to a cloud end;
after receiving the vibration acceleration and the surface temperature data, the cloud converts the three-axis vibration acceleration from a time domain to a frequency domain through Fourier transform;
inputting the converted signals into a deep learning network for detection, wherein the number of input signal channels is four;
taking the deep features extracted by the resnet18 network as input, and acquiring a learning dependence problem in signal sequence data through ON-LSTM;
the characteristics extracted by the ON-LSTM network are used as input, and the SVM receives the input signal, trains and outputs the information expressed by the signal, the normal signal and various fault signals to obtain the specific fault type of the signal.
2. The bearing fault diagnosis method based on deep learning of claim 1, wherein: the portable vibration signal collector is used for collecting the vibration acceleration of X, Y and Z three axes of the industrial field mechanical equipment and the surface temperature data of the equipment through 5G edge calculation so as to collect the equipment bearing in operation.
3. The deep learning-based bearing fault diagnosis method according to claim 1, characterized in that: and inputting the converted signals into a deep learning network for detection, wherein the number of input signal channels is four, and the input signal channels are respectively vibration acceleration of three axes X, Y and Z and temperature signals of the surface of the equipment.
4. A deep learning based bearing fault diagnosis system comprising:
the signal acquisition module: the portable vibration signal collector is used for collecting vibration acceleration of X, Y and Z three axes of industrial field mechanical equipment and equipment surface temperature data through a 5G edge calculation portable vibration signal collector;
a data transmission module: the system is used for transmitting the vibration acceleration of the X, Y and Z three axes of the industrial field mechanical equipment and the surface temperature data of the equipment to the cloud end;
a deep learning module: the cloud end is used for receiving the vibration acceleration and the surface temperature data and then converting the three-axis vibration acceleration from a time domain to a frequency domain through Fourier transform;
inputting the converted signals into a deep learning network for detection, wherein the number of input signal channels is four;
taking the deep features extracted by the resnet18 network as input, and acquiring a learning dependence problem in signal sequence data through ON-LSTM;
a fault type output module: the method is used for taking the time domain characteristics extracted by the ON-LSTM network as input, and outputting information expressed by the signal, normal signals and various fault signals after the support vector machine SVM receives the input signal for training to obtain the specific fault type of the signal.
5. The bearing fault diagnosis system of claim 4, wherein the portable vibration signal collector for 5G edge computing collects vibration acceleration and device surface temperature data of three axes X, Y and Z of the industrial field mechanical device so as to collect the device bearing in operation.
6. The bearing fault diagnosis system according to claim 4, wherein the converted signals are input into a deep learning network for detection, and the number of input signal channels is four, namely the vibration acceleration of three axes X, Y and Z and the temperature signal of the surface of the equipment.
7. A computer-readable storage medium, wherein the computer-readable storage medium comprises: a stored program, wherein the program when executed performs the method of any one of claims 1 to 3.
8. An electronic device comprising a memory and a processor, wherein the memory has stored therein a computer program and the processor is arranged to execute the method according to any of claims 1-3 by means of the computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211147956.5A CN115374829A (en) | 2022-09-20 | 2022-09-20 | Deep learning-based bearing fault diagnosis method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211147956.5A CN115374829A (en) | 2022-09-20 | 2022-09-20 | Deep learning-based bearing fault diagnosis method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115374829A true CN115374829A (en) | 2022-11-22 |
Family
ID=84071464
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211147956.5A Pending CN115374829A (en) | 2022-09-20 | 2022-09-20 | Deep learning-based bearing fault diagnosis method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115374829A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117516927A (en) * | 2024-01-05 | 2024-02-06 | 四川省机械研究设计院(集团)有限公司 | Gearbox fault detection method, system, equipment and storage medium |
-
2022
- 2022-09-20 CN CN202211147956.5A patent/CN115374829A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117516927A (en) * | 2024-01-05 | 2024-02-06 | 四川省机械研究设计院(集团)有限公司 | Gearbox fault detection method, system, equipment and storage medium |
| CN117516927B (en) * | 2024-01-05 | 2024-04-05 | 四川省机械研究设计院(集团)有限公司 | Gearbox fault detection method, system, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109635677B (en) | Compound fault diagnosis method and device based on multi-label classification convolutional neural network | |
| CN113076834B (en) | Rotating machine fault information processing method, processing system, processing terminal, and medium | |
| CN112113755B (en) | Mechanical fault intelligent diagnosis method based on deep convolution-kurtosis neural network | |
| CN113505664B (en) | Fault diagnosis method for planetary gear box of wind turbine generator | |
| CN111650453A (en) | Power equipment diagnosis method and system based on windowing characteristic Hilbert imaging | |
| CN111931389A (en) | Method and device for analyzing normal and abnormal running state of rotary equipment | |
| CN113339204B (en) | Wind driven generator fault identification method based on hybrid neural network | |
| Li et al. | Joint attention feature transfer network for gearbox fault diagnosis with imbalanced data | |
| CN111562105B (en) | Wind turbine generator gearbox fault diagnosis method based on wavelet packet decomposition and convolutional neural network | |
| CN116593157A (en) | Complex working condition gear fault diagnosis method based on matching element learning under small sample | |
| CN111795819B (en) | Gear box fault diagnosis method integrating vibration and current signal collaborative learning | |
| CN115374829A (en) | Deep learning-based bearing fault diagnosis method and system | |
| CN116956215A (en) | Fault diagnosis method and system for transmission system | |
| Wang et al. | Fault diagnosis of industrial robots based on multi-sensor information fusion and 1D convolutional neural network | |
| CN103758742A (en) | Plunger pump failure analysis system based on double-class feature fusion diagnosis | |
| CN111539381B (en) | Construction method of wind turbine bearing fault classification diagnosis model | |
| CN117150340A (en) | Method and device for diagnosing faults of small samples of switch machine | |
| CN113268552B (en) | Generator equipment hidden danger early warning method based on locality sensitive hashing | |
| CN114065651A (en) | Fault time prediction method for rotary equipment | |
| CN113723592A (en) | Fault diagnosis method based on wind power gear box monitoring system | |
| CN114442543A (en) | Computer monitoring method suitable for early warning of hydropower station fault | |
| CN112269778A (en) | Equipment fault diagnosis method | |
| CN111476383A (en) | Pump station unit state maintenance dynamic decision method | |
| Zou et al. | A Bayesian Adaptive Resize-Residual Deep Learning Network for Fault Diagnosis of Rotating Machinery | |
| CN116973794B (en) | Lithium battery SOH estimation method based on incomplete charging voltage curve reconstruction |
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 |