CN116402825B - Bearing fault infrared diagnosis method, system, electronic equipment and storage medium - Google Patents
Bearing fault infrared diagnosis method, system, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN116402825B CN116402825B CN202310677567.1A CN202310677567A CN116402825B CN 116402825 B CN116402825 B CN 116402825B CN 202310677567 A CN202310677567 A CN 202310677567A CN 116402825 B CN116402825 B CN 116402825B
- Authority
- CN
- China
- Prior art keywords
- bearing
- bls
- infrared
- features
- optimal
- 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.)
- Active
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 78
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000003860 storage Methods 0.000 title claims abstract description 12
- 239000011159 matrix material Substances 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims abstract description 29
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 23
- 230000002068 genetic effect Effects 0.000 claims abstract description 23
- 238000012360 testing method Methods 0.000 claims abstract description 15
- 238000005457 optimization Methods 0.000 claims abstract description 13
- 230000008859 change Effects 0.000 claims abstract description 10
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 230000008569 process Effects 0.000 claims description 20
- 238000005096 rolling process Methods 0.000 claims description 15
- 239000013598 vector Substances 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 11
- 238000006243 chemical reaction Methods 0.000 claims description 10
- 238000012216 screening Methods 0.000 claims description 8
- 238000009826 distribution Methods 0.000 claims description 7
- 238000002474 experimental method Methods 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000002405 diagnostic procedure Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 6
- 238000010353 genetic engineering Methods 0.000 claims description 4
- 238000005520 cutting process Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 230000036541 health Effects 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims 1
- 238000004891 communication Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 9
- 239000010410 layer Substances 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 238000013135 deep learning Methods 0.000 description 5
- 239000004973 liquid crystal related substance Substances 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 238000013507 mapping Methods 0.000 description 4
- 230000035772 mutation Effects 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 235000002566 Capsicum Nutrition 0.000 description 1
- 239000006002 Pepper Substances 0.000 description 1
- 235000016761 Piper aduncum Nutrition 0.000 description 1
- 235000017804 Piper guineense Nutrition 0.000 description 1
- 244000203593 Piper nigrum Species 0.000 description 1
- 235000008184 Piper nigrum Nutrition 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000005314 correlation function Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 150000003839 salts Chemical class 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 239000002356 single layer Substances 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
- 238000003892 spreading Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Quality & Reliability (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Image Analysis (AREA)
Abstract
The application provides a bearing fault infrared diagnosis method, a system, electronic equipment and a storage medium, belonging to the field of bearing fault diagnosis; the method comprises the steps of collecting multi-frame infrared images; acquiring an interested picture according to a temperature change area in the training set image; preprocessing the picture of interest to convert the picture into a binary image; extracting second-order statistical features from the binary image based on a gray level co-occurrence matrix formula; performing feature iterative optimization on the second-order statistical features to obtain optimal features; optimizing the BLS network parameters through a genetic algorithm to obtain optimal parameters; training a BLS network based on the optimal characteristics and the optimal parameters to obtain a CFS-GA-BLS fault diagnosis model; inputting the test set image into a CFS-GA-BLS fault diagnosis model to output a fault infrared diagnosis result of the bearing to be diagnosed. The application can improve the high efficiency and high accuracy of the infrared diagnosis result of bearing faults.
Description
Technical Field
The application belongs to the technical field of bearing fault diagnosis, and particularly relates to a bearing fault infrared diagnosis method, a bearing fault infrared diagnosis system, electronic equipment and a storage medium.
Background
Rolling bearings are an indispensable important mechanical element, but they often work for a long period of time under high loads and complex environments, and performance is degraded and the bearings are disabled. During this process, various failures such as wear, corrosion, cracks, etc. tend to occur. Therefore, effective bearing failure diagnosis is critical to maintaining safe operation of the machinery. The conventional bearing fault diagnosis is mostly based on vibration signals and acoustic emission signals. However, both of these signal acquisitions are susceptible to the effects of the external environment, i.e., noise is easily acquired, making it difficult to extract useful information therefrom. In recent years, temperature signals have the characteristics of non-contact, high sensitivity and strong real-time performance, and have strong advantages for test targets running at high speed, and some students introduce the temperature signals into fault diagnosis of mechanical equipment. The temperature signal may be captured by a thermal infrared imager to display thermal characteristics of the object. Compared with two signals of a vibration signal and an acoustic emission signal, the infrared image has certain advantages as an object of bearing fault diagnosis.
Currently, in bearing failure diagnosis, methods of shallow machine learning and deep learning are generally used. Shallow machine learning methods have poor prediction generalization capability and insufficient precision, deep learning networks require large amounts of data, training time is too long, and are often regarded as a black box, and meaning interpretation is very complex. As the BLS network is proposed, the calculating speed and measuring accuracy of the model can be improved comprehensively by calculating the weights of the feature nodes and the enhancement nodes through pseudo-inverse, and the BLS network has the greatest characteristic that the BLS network only comprises a single hidden layer, and when the BLS network faces the inaccurate learning situation, incremental expansion is performed in a transverse mode at any time, so that the problem of overlong deep learning training time is solved greatly. However, in the field of bearing fault diagnosis, no related research is performed on optimizing model parameters and infrared image characteristics of a BLS network, and in the process of completing bearing fault diagnosis, the diagnosis rate and the diagnosis accuracy are often affected by selection of the infrared image characteristics and the model parameters.
Therefore, it is important to provide an infrared diagnosis method for bearing faults with high efficiency and high precision.
Disclosure of Invention
In order to solve the technical problems, the application provides a bearing fault infrared diagnosis method, a system, electronic equipment and a storage medium, which can improve the high efficiency and high accuracy of bearing fault infrared diagnosis results.
In a first aspect, the present application provides an infrared bearing fault diagnosis method, comprising:
collecting multi-frame infrared images of the running state of a bearing to be diagnosed of a preset type through a rolling bearing fault experiment table; the multi-frame infrared image is divided into a training set image and a testing set image;
cutting according to a temperature change area in the training set image to obtain an interested picture;
preprocessing the picture of interest to convert the picture of interest into a binary image;
extracting second-order statistical features from the binary image based on a gray level co-occurrence matrix formula; the second-order statistical features comprise energy, contrast, entropy, mean value, variance and correlation;
performing feature iterative optimization on the second-order statistical features to obtain optimal features;
optimizing the BLS network parameters through a genetic algorithm to obtain optimal parameters; the BLS network parameters comprise the number N1 of each window characteristic node, the number N2 of windows of the characteristic nodes and the number N3 of enhancement nodes;
Training the BLS network based on the optimal characteristics and the optimal parameters to obtain a CFS-GA-BLS fault diagnosis model;
and inputting the test set image into the CFS-GA-BLS fault diagnosis model so as to output a fault infrared diagnosis result of the bearing to be diagnosed.
Preferably, the step of converting the picture of interest into a binary image by preprocessing specifically includes:
graying the picture of interest;
carrying out median filtering denoising treatment on the interested picture subjected to graying treatment by adopting a filter with the size of 3 multiplied by 3;
calculating local mean and standard deviation of the image according to brightness distribution of different areas of the image of interest after denoising treatment, and calculating a local threshold value based on the local mean and standard deviation;
obtaining an optimal threshold value by adopting a Gaussian weighted average algorithm based on the local threshold value;
and converting the interested picture into a binary image according to the optimal threshold value.
Preferably, the step of performing feature iterative optimization for the second-order statistical feature to obtain an optimal feature specifically includes:
calculating correlation coefficients among the second-order statistical features by adopting CFS, and generating a correlation coefficient matrix;
expanding the correlation coefficient matrix into a one-dimensional vector by using stack operation;
Selecting a target feature with the largest occurrence frequency from the features with the highest correlation based on the sequence of the one-dimensional vectors;
the target features are ranked at the forefront and are used as the optimal features of the second-order statistical features.
Preferably, the step of optimizing the BLS network parameter by the genetic algorithm to obtain an optimal parameter specifically includes:
setting parameter ranges of N1, N2 and N3, and generating an initialization population;
selecting a reserve dominant individual based on the advanced fitness of the initialized population;
subjecting the remaining dominant individual to a series of genetic manipulations to derive progeny;
and screening out optimal parameters meeting preset targets from the offspring.
In a second aspect, the present application provides an infrared bearing failure diagnostic system comprising:
the acquisition module is used for acquiring multi-frame infrared images of the running state of the bearing to be diagnosed of a preset type through the rolling bearing fault experiment table; the multi-frame infrared image is divided into a training set image and a testing set image;
the clipping module is used for clipping according to the temperature change area appearing in the training set image to obtain an interested picture;
the conversion module is used for preprocessing the interested picture and converting the interested picture into a binary image;
The extraction module is used for extracting second-order statistical features from the binary image based on a gray level co-occurrence matrix formula; the second-order statistical features comprise energy, contrast, entropy, mean value, variance and correlation;
the iteration module is used for carrying out characteristic iteration optimization aiming at the second-order statistical characteristics to obtain optimal characteristics;
the genetic module is used for optimizing the BLS network parameters through a genetic algorithm to obtain optimal parameters; the BLS network parameters comprise the number N1 of each window characteristic node, the number N2 of windows of the characteristic nodes and the number N3 of enhancement nodes;
the training module is used for training the BLS network based on the optimal characteristics and the optimal parameters to obtain a CFS-GA-BLS fault diagnosis model;
and the diagnosis module is used for inputting the test set image into the CFS-GA-BLS fault diagnosis model so as to output the fault infrared diagnosis result of the bearing to be diagnosed.
Preferably, the conversion module specifically includes:
the gray level unit is used for carrying out gray level processing on the picture of interest;
the denoising unit is used for carrying out median filtering denoising treatment on the image of interest after the graying treatment by adopting a filter with the size of 3 multiplied by 3;
The computing unit is used for computing local mean values and standard deviations of the images according to brightness distribution of different areas of the interested images after denoising, and computing local threshold values based on the local mean values and the standard deviations;
the weighting unit is used for obtaining an optimal threshold value by adopting a Gaussian weighted average algorithm based on the local threshold value;
and the conversion unit is used for converting the picture of interest into a binary image according to the optimal threshold value.
Preferably, the iteration module specifically includes:
the generation unit is used for calculating the correlation coefficient between the second-order statistical features by adopting CFS and generating a correlation coefficient matrix;
the expanding unit is used for expanding the correlation coefficient matrix into a one-dimensional vector by using stack operation;
a selecting unit, configured to select a target feature with the largest occurrence number from features with the highest correlation based on the ranking of the one-dimensional vectors;
and the sorting unit is used for sorting the target feature at the forefront and taking the target feature as the optimal feature of the second-order statistical feature.
Preferably, the genetic module specifically includes:
the setting unit is used for setting the parameter ranges of N1, N2 and N3 and generating an initialization population;
a selection unit for selecting a reserve dominant individual based on the advanced fitness of the initialized population;
A genetic unit for deriving progeny from a series of genetic manipulations of said dominant individual remaining;
and the screening unit is used for screening out the optimal parameters meeting the preset targets from the offspring.
In a third aspect, the present application provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the bearing failure infrared diagnostic method according to the first aspect when executing the computer program.
In a fourth aspect, the present application provides a storage medium having stored thereon a computer program which, when executed by a processor, implements the bearing failure infrared diagnostic method according to the first aspect.
Compared with the prior art, the bearing fault infrared diagnosis method, the system, the electronic equipment and the storage medium provided by the application have the following beneficial effects:
1. the temperature signal of the infrared image is adopted to replace the original vibration signal and sound emission signal, so that the fault is in a visual state, noise in the signal acquisition process is avoided, and the instantaneity and the accuracy of the acquired image are improved.
2. The second-order statistical features comprising 72 features including distance and angle energy, contrast, entropy, mean value, variance, correlation and the like are extracted through the gray level co-occurrence matrix, and the optimal feature number and the corresponding feature are searched in an iterative mode, so that the problem of difficulty in feature selection in the existing bearing fault diagnosis method is solved.
3. The genetic algorithm is used for optimizing the BLS network parameters to obtain the optimal parameters, so that the optimal parameters and the highest accuracy can be obtained, and the problem that the BLS network parameters are difficult to set in the existing bearing fault diagnosis method is solved.
4. Based on the optimal characteristics and optimal parameters, training the BLS network to obtain a CFS-GA-BLS fault diagnosis model, and through the CFS-GA-BLS fault diagnosis model, the efficiency and the precision of bearing fault infrared diagnosis can be improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of the infrared diagnosis method for bearing failure provided in embodiment 1 of the present invention;
FIG. 2 is a schematic view of an infrared image of each type of rolling bearing according to embodiment 1 of the present invention;
fig. 3 is a schematic view of adaptive threshold segmentation of images of interest for each type of rolling bearing according to embodiment 1 of the present invention;
Fig. 4 is a diagnosis result of the CFS-GA-BLS fault diagnosis model provided in example 1 of the present invention.
FIG. 5 is a block diagram of a bearing failure infrared diagnostic system corresponding to the method of embodiment 1 provided in embodiment 2 of the present invention;
fig. 6 is a schematic hardware structure of an electronic device according to embodiment 3 of the present invention.
Reference numerals illustrate:
10-an acquisition module;
20-cutting module;
a 30-conversion module, a 31-gray level unit, a 32-denoising unit, a 33-calculation unit, a 34-weighting unit and a 35-conversion unit;
40-an extraction module;
the device comprises a 50-iteration module, a 51-generation unit, a 52-expansion unit, a 53-selection unit and a 54-ordering unit;
60-genetic module, 61-setting unit, 62-selection unit, 63-genetic unit, 64-screening unit;
70-a training module;
80-a diagnostic module;
90-bus, 91-processor, 92-memory, 93-communication interface.
Detailed Description
Example embodiments will now be described more fully with reference to the accompanying drawings. However, the exemplary embodiments may be embodied in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art.
Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments of the disclosure. One skilled in the relevant art will recognize, however, that the disclosed aspects may be practiced without one or more of the specific details, or with other methods, components, devices, steps, etc. In other instances, well-known methods, devices, implementations, or operations are not shown or described in detail to avoid obscuring aspects of the disclosure.
The block diagrams depicted in the figures are merely functional entities and do not necessarily correspond to physically separate entities. That is, the functional entities may be implemented in software, or in one or more hardware modules or integrated circuits, or in different networks and/or processor devices and/or microcontroller devices.
The flow diagrams depicted in the figures are exemplary only, and do not necessarily include all of the elements and operations/steps, nor must they be performed in the order described. For example, some operations/steps may be decomposed, and some operations/steps may be combined or partially combined, so that the order of actual execution may be changed according to actual situations.
Machine learning and deep learning are heavily cited to infrared image fault diagnosis of mechanical or electrical power. However, machine learning often has data dependence and the generalization ability of models is poor. The depth model requires a lot of time and does not have good interpretation. Therefore, the fault diagnosis using the above two modes has a certain defect. In 2018, a width learning network is provided, which is a random vector function linked neural network, and the computing speed and the measuring precision of a model can be comprehensively improved by calculating the weights of characteristic nodes and enhancement nodes through pseudo-inverse, and the BLS network is characterized by only comprising a hidden layer of a single layer, and when the situation of inaccurate learning is faced, incremental expansion is performed in a transverse mode at any time, so that the problem of overlong deep learning training time is greatly solved. Wang et al have completed extracting features describing temporal variations and spatial distributions from a sequence of infrared images of electrical equipment, which are ultimately diagnosed with BLS. Zhang et al use dual-tree complex wavelet to decompose the vibration signal of bearing fault into sub-signals of different frequency bands, then extract the sub-bands as feature vectors, and finally train the samples with BLS to accomplish fast fault classification. Therefore, the width learning is introduced into the field of fault diagnosis to have a certain practical meaning. However, in the field of bearing fault diagnosis, no related research is performed on optimizing model parameters and infrared image characteristics of a BLS network, and in the process of completing bearing fault diagnosis, the diagnosis rate and the diagnosis accuracy are often affected by selection of the infrared image characteristics and the model parameters. The present application has been made in view of this.
Example 1
Specifically, fig. 1 is a schematic flow chart of an infrared diagnosis method for bearing faults provided in the present embodiment.
As shown in fig. 1, the bearing fault infrared diagnosis method of the present embodiment includes the following steps:
s101, collecting multi-frame infrared images of the running state of the bearing to be diagnosed of a preset type through a rolling bearing fault experiment table.
The multi-frame infrared image is divided into a training set image and a testing set image.
In this embodiment, the preset types of the running states of the bearing to be diagnosed include health, cage fault, inner ring 0.5mm fault, inner ring 1.0mm fault, inner ring 1.5mm fault, outer ring 0.5mm fault, outer ring 1.0mm fault, outer ring 1.5mm fault, rolling element pitting fault. Specifically, the rolling bearing fault experiment table applied in the embodiment comprises a servo motor, a controller thereof, a coupler, a shaft, a disc and a shaft, a fault bearing and an infrared thermal image acquisition system, wherein the type of the bearing is S6205-2RSR, and the motor speed is stabilized at 2000rmp. The infrared thermal image acquisition system comprises BM_RI Software V7.4 and an infrared thermal imager FLIRA35, and parameters of the infrared thermal imager and the acquisition system are shown in the following table.
In the specific acquisition process, the ambient temperature is kept at about 24.5 ℃, and when video acquisition is carried out on each bearing state, the temperature is kept stable. An infrared video was then acquired for 15 minutes. The acquisition steps of the infrared thermal image are as follows:
step 1: a bearing is selected for mounting to the shaft.
Step 2: and opening the motor controller to ensure that the bearing operates at a constant speed of 2000 rmp.
Step 3: and observing the temperature change, and starting to acquire infrared video for 15 minutes after the temperature change is approximately stable.
Step 4: the rolling bearing failure bench was cooled to ambient temperature 24.5 ℃.
Step 6: repeating the steps 1-4, and collecting 9 different bearing fault infrared videos for 15 minutes.
Step 7: at 9s intervals, infrared images were extracted from the video, 100 sheets per category, for a total of 900 sheets. Taking out 80 sheets of each category, and taking a total of 800 sheets as training set images; a total of 10 sheets were taken out for each category, and 100 sheets were taken as test set images.
S102, clipping is carried out according to the temperature change area appearing in the training set image to obtain an interested picture.
Specifically, a bearing and shaft region generating temperature change is selected from 800 training set images, and clipping is performed to obtain an interested picture. A picture of interest of the infrared image extracted by each type of bearing is shown in fig. 2; IN fig. 2, the label "HE" indicates a healthy bearing, the label "HO" indicates a cage failure bearing, the label "IN05" indicates an inner ring 0.5mm failure bearing, the label "IN10" indicates an inner ring 1.0mm failure bearing, the label "IN15" indicates an inner ring 1.5mm failure bearing, the label "OU05" indicates an outer ring 0.5mm failure bearing, the label "OU10" indicates an outer ring 1.0mm failure bearing, the label "OU15" indicates an outer ring 1.5mm failure bearing, and the label "RO" indicates a rolling element pitting failure bearing.
S103, preprocessing the interested picture to convert the interested picture into a binary image.
In particular, infrared thermal images tend to contain more complex, more intense noise than ordinary optical images, subject to factors such as space-time environment and detection equipment. Also, the temperature of the surrounding environment may interfere with the key information of the picture. Therefore, preprocessing work for the picture of interest is necessary.
Further, the specific steps of step S103 include:
s1031, carrying out gray scale processing on the interested picture.
Specifically, the image of interest is converted from an RGB true color image to a two-dimensional gray image, and the conversion formula is as follows:
；
in the method, in the process of the invention,Grayfor the converted gray pixel value,Rfor the red channel pixel value,Gis the pixel value of the green channel,Bis the blue pixel value.
S1032, median filtering denoising is carried out on the interested picture after the graying treatment by adopting a filter with the size of 3 multiplied by 3.
Specifically, noise caused by working environment, transmission medium and material properties in the gray level picture, such as salt and pepper noise generated by image shooting and normally distributed Gaussian noise, is removed. And setting the gray value of each pixel point to be the median value of the gray values of all the pixel points in a certain neighborhood window of the point by adopting a median filtering denoising method. The specific formula is as follows:
；
In the method, in the process of the invention,g(. Cndot.) is the original pixel value,f(. Cndot.) is the pixel value after processing,Mfor the number of surrounding pixels to be taken,xandyis the pixel coordinates.
S1033, calculating local mean and standard deviation according to brightness distribution of different areas of the interested picture after denoising, and calculating a local threshold value based on the local mean and standard deviation.
Specifically, the difference caused by the difference of the background temperatures is processed by adopting the adaptive threshold segmentation, the size of each local area is set to be 11×11 window, and the local mean value and the local standard deviation of each local area are calculated.
And S1034, obtaining an optimal threshold value by adopting a Gaussian weighted average algorithm based on the local threshold value.
Specifically, the weight of a pixel in the neighborhood of the gaussian weighted average algorithm in this embodiment is a gaussian function of the gray level difference between the pixel and the center pixel. The gaussian curve shape is very similar to the curve shape of a general correlation function, so that it is a more ideal weighting characteristic curve.
S1035, converting the interested picture into a binary image according to the optimal threshold.
Specifically, a local threshold is calculated according to the local mean value and the local standard deviation, and an optimal threshold is calculated by adopting a Gaussian weighted average algorithm, so that a binarized image is obtained, a final binarized image is formed, and a result after self-adaptive threshold segmentation is shown in fig. 3.
S104, extracting second-order statistical features from the binary image based on a gray level co-occurrence matrix formula.
Wherein the second order statistical features include energy, contrast, entropy, mean, variance, and correlation. Specifically, the gray level co-occurrence matrix formula is specifically as follows:
；
in the method, in the process of the invention,P(i,j,δ,θ) Representing gray values asiAndjis at a distance from each otherδAnd angle ofθIs used to determine the probability of the occurrence of a location of (c),N(i,j,δ,θ) Representing gray values asiAndjis at a distance from each otherδAnd angle ofθIs provided with a number of positions in the pattern,Gthe gray level number of the binary image, i.e., 256.
The above formula selects pixel distances with distances of 1, 2 and 3, and selects pixel angles of 0 °, 45 °, 90 ° and 135 °, and the extracted second-order statistical features include energy, contrast, entropy, mean, variance and correlation, and total 72 features are respectively named as glcm_0, glcm_1, … and glcm_71. Wherein the energy isf 1 Said contrast ratiof 2 Said entropyf 3 The mean value off 4 Said variance isf 5 The correlation isf 6 The calculation formulas of (a) are respectively as follows:
；/>；/>；
；/>；/>；
in the method, in the process of the invention,N g representing the number of different gray levels in the binary image,p（i,j) Representing the first degree in the normalized gray level co-occurrence matrixi,j) The number of objects to be achieved in the process,nrepresents the number of the row of the codes,μ x 、μ y 、σ x andσ y representation ofp x （i) And (3) withp y （j) Is defined as the mean value and standard deviation of (c), μIs thatμ x Andμ y average value of (2). Wherein, the liquid crystal display device comprises a liquid crystal display device,f 6 in the calculation formula of (a),
in the method, in the process of the invention,p x （i) Represents the probability of a boundary in the lateral direction,p y （j) Representing the longitudinal boundary probability.
Specifically, energyf 1 Representing the gray level distribution uniformity degree and the texture thickness degree of the image; contrast ratiof 2 Characterizing the definition and the texture depth of the image; entropy off 3 A metric characterizing the random amount of the image, representing the complexity of the image; mean value off 4 Characterizing the degree of regularity of the texture; variance off 5 A measure characterizing the deviation of the pixel values from the mean; correlation off 6 And the similarity degree in the element direction in the symbiotic matrix is represented and measured.
S105, performing feature iterative optimization on the second-order statistical features to obtain optimal features.
Specifically, the second-order statistical features including 72 features including distance and angle energy, contrast, entropy, mean value, variance, correlation and the like are extracted through the gray level co-occurrence matrix, the optimal feature number is searched in an iterative mode, and the optimal feature number and the corresponding feature are arranged side by side, so that the problem of difficulty in feature selection in the existing bearing fault diagnosis method is solved. As the characteristic and the target variable have stronger association relation, the classifier performance can be improved. The robust features can reduce the influence of abnormal data values and noise, and improve the stability of the classifier, and the embodiment adopts a corelation-based Feature Selection (CFS) method, which utilizes the robustness to sort the features according to a certain weight, so that the forefront features are preferentially extracted.
Further, the specific steps of step S105 include:
s1051, calculating the correlation coefficient between the second-order statistical features by adopting CFS, and generating a correlation coefficient matrix.
Specifically, the correlation coefficient according to the present embodiment employs a pearson correlation coefficient.
S1052, expanding the correlation coefficient matrix into a one-dimensional vector by using a stack operation.
Specifically, calculating the absolute value of a correlation coefficient matrix to ensure that the values of all the correlation coefficients are positive numbers; the elements on the diagonal are culled, as they represent the correlation of the variable with itself, always being 1. The matrix of correlation numbers is expanded into a one-dimensional array such that each element represents a correlation coefficient.
S1053, selecting the target feature with the largest occurrence number from the features with the highest correlation based on the sequence of the one-dimensional vectors.
Specifically, for the row to which each element belongs, it is compared with the correlation coefficients of other variables. And according to the comparison result, reserving the maximum correlation coefficient. The above operation is repeated for each variable until the required number of correlation coefficients are obtained.
S1054, the target feature is ranked at the forefront, and is used as the optimal feature of the second-order statistical feature.
S106, optimizing the BLS network parameters through a genetic algorithm to obtain optimal parameters.
The BLS network parameters comprise the number N1 of each window characteristic node, the number N2 of windows of the characteristic nodes and the number N3 of enhancement nodes.
Specifically, the genetic algorithm is used for optimizing the BLS network parameters to obtain the optimal parameters, so that the optimal parameters and the highest accuracy can be obtained, and the problem that the BLS network parameters are difficult to set in the existing bearing fault diagnosis method is solved. In specific practice, the Genetic Algorithm (GA) is an optimization algorithm, following the principles of survival and disfavored. It is developed by referring to the genetic process in evolutionary biology, and comprises a series of operations of inheritance, mutation, crossover, natural selection and the like.
Further, the specific steps of step S106 include:
s1061, setting the parameter ranges of N1, N2 and N3, and generating an initialization population.
Specifically, the BLS network is a random vector function linked neural network, and can comprehensively improve the operation speed and measurement accuracy of the model by calculating weights of the feature nodes and the enhancement nodes through pseudo-inversion. The BLS network operates as follows:
first, the model completes the mapping of the input data to the feature nodes, as follows:
；
Wherein Z is i Representing characteristic nodeiThe number of the two-dimensional space-saving type,Xthe characteristics of the input are represented by a representation,W ei representing the random weights of the feature map layer,β ei representing the feature map layer random bias,ψ(-) represents a mapping function of data X;
secondly, generating an enhancement node by mapping the node through transformation, wherein the formula is as follows:
；
wherein H is j Representing enhanced node numberjThe number of the two-dimensional space-saving type,representing a set of mapped nodes, W hj Representing feature map layer random weights, beta hj Representing feature mapping layer random bias, ζ j (. Cndot.) means each Z i Is a projection function of (2);
again, all the mapped nodes and enhanced nodes are output, as follows:
；
wherein Y represents a node set, Z n Representing a set of mapped nodes, H m Representing an enhanced node set, W m Representing a set of weights;
finally, calculating the pseudo-inverse to obtain the weight of the output layer, wherein the formula is as follows:
；
therefore, it is very important to set the number of feature nodes, the number of feature node windows, and the number of enhancement nodes in the entire network model, which are denoted as N1, N2, and N3, respectively. Typically, these parameters require manual adjustment to bring the diagnostic result to a high level. However, such models do not have generalization capability, nor do they necessarily guarantee that the optimal diagnostic rate is obtained. Therefore, parameter optimization for N1, N2 and N3 is very necessary.
And S1062, selecting dominant individuals to be preserved based on the advanced fitness of the initialized population.
Specifically, the fault diagnosis accuracy of the BLS model is used as a fitness function, and N1, N2, and N3 are used as optimization parameters. After setting the parameter ranges of N1, N2 and N3, generating an initial population. Initial population advancement fitness, and selection of dominant individuals for retention.
S1063, carrying out series of genetic operations on the reserved dominant individual to derive offspring.
Specifically, a genetic algorithm is used for optimizing the BLS network, an initial population is set, N1, N2 and N3 are used as optimizing parameters, parameter selection ranges are respectively set, BLS is used as an objective function, and therefore optimal solutions and optimal parameters of N1, N2 and N3 are found. In the model, the shrinkage factor of the BLS network is set to 0.7, the regularization factor is set to 2-20, the N1 setting range is [10, 30], the N2 setting range is [10, 30], the N3 setting range is [500, 600], the population number initial value of the GA algorithm is set to 30, the crossover probability is set to 0.8, the mutation probability is set to 0.1, and the number of iterations is set to 10.
S1064, screening out optimal parameters meeting preset targets from the offspring.
Specifically, an initial population is generated for the parameters to be optimized through the genetic algorithm, the initial population is evaluated according to the fitness function, and a solution set with a smaller fitness function is eliminated. The remained population with high fitness is subjected to operations such as amplitude, intersection, mutation and the like, and individuals are continuously optimized, so that an optimal solution is obtained.
And S107, training the BLS network based on the optimal characteristics and the optimal parameters to obtain a CFS-GA-BLS fault diagnosis model.
Specifically, the CFS sequentially invokes the n features ordered first, the 1 st, the 2 nd, the first 2, the 72 nd, the first 72, so fewer and more representative features can be selected. The 72 sets of features are then sequentially input into the GA-BLS and cycled 72 times, each round corresponding to the output accuracy and corresponding parameters.
S108, inputting the test set image into the CFS-GA-BLS fault diagnosis model so as to output a fault infrared diagnosis result of the bearing to be diagnosed.
It should be noted that, according to the output result of the CFS-GA-BLS fault diagnosis model, the final GA-BLS fault diagnosis accuracy of each feature selection is plotted, and the result is shown in fig. 4. As features continue to increase, fault diagnosis accuracy continues to increase until features are selected to 49, with the highest accuracy 0.98889, and then higher accuracy is not achieved. In the whole feature optimization process, the accuracy is greatly reduced due to the fact that features are added several times, the relevance between the added features and other features is small, but the relevance between the features is large, and therefore the whole representative features are disturbed. In summary, CFS-GA-BLS selected the best 42 features, with N1, N2, N3 having the best values of 21, 12, 541 and the highest accuracy of 0.98889.
Example 2
This embodiment provides a block diagram of a system corresponding to the method described in embodiment 1. Fig. 5 is a block diagram of a bearing failure infrared diagnostic system according to the present embodiment, as shown in fig. 5, including:
the acquisition module 10 is used for acquiring multi-frame infrared images of the running state of the bearing to be diagnosed of a preset type through a rolling bearing fault experiment table; the multi-frame infrared image is divided into a training set image and a testing set image;
specifically, the preset types of the running states of the bearing to be diagnosed comprise health, cage faults, inner ring 0.5mm faults, inner ring 1.0mm faults, inner ring 1.5mm faults, outer ring 0.5mm faults, outer ring 1.0mm faults, outer ring 1.5mm faults and rolling element pitting faults.
And the clipping module 20 is used for clipping according to the temperature change area appearing in the training set image to obtain the picture of interest.
The conversion module 30 is configured to perform preprocessing on the picture of interest and convert the picture of interest into a binary image.
An extracting module 40, configured to extract second-order statistical features from the binary image based on a gray level co-occurrence matrix formula; the second-order statistical features comprise energy, contrast, entropy, mean value, variance and correlation;
Specifically, the gray level co-occurrence matrix formula is specifically as follows:
；
in the method, in the process of the invention,P(i,j,δ,θ) Representing gray values asiAndjis at a distance from each otherδAnd angle ofθIs used to determine the probability of the occurrence of a location of (c),N(i,j,δ,θ) Representing gray values asiAndjis at a distance from each otherδAnd angle ofθIs provided with a number of positions in the pattern,Ga gray level number representing a binary image, i.e., 256;
further, each second-order statistical feature is a set of 72 features in total, wherein the pixel distance is 1, 2 and 3, the angle is 0, 45, 90 and 135 respectively, and the energy, contrast, entropy, mean value, variance and correlation are all set; said energyf 1 Said contrast ratiof 2 Said entropyf 3 The mean value off 4 Said variance isf 5 The correlation isf 6 The calculation formulas of (a) are respectively as follows:
；/>；/>；
；/>；/>；
in the method, in the process of the invention,N g representing the number of different gray levels in the binary image,p（i,j) Representing the first degree in the normalized gray level co-occurrence matrixi,j) The number of objects to be achieved in the process,nrepresents the number of the row of the codes,μ x 、μ y 、σ x andσ y representation ofp x （i) And (3) withp y （j) Is defined as the mean value and standard deviation of (c),μis thatμ x Andμ y average value of (2).Wherein, the liquid crystal display device comprises a liquid crystal display device,f 6 in the calculation formula of (a),
/>
in the method, in the process of the invention,P x （i) Represents the probability of a boundary in the lateral direction,p y （j) Representing the longitudinal boundary probability.
And the iteration module 50 is used for carrying out feature iteration optimization aiming at the second-order statistical features to obtain optimal features.
A genetic module 60, configured to obtain optimal parameters for the BLS network parameters by performing optimization through a genetic algorithm; the BLS network parameters comprise the number N1 of each window characteristic node, the number N2 of windows of the characteristic nodes and the number N3 of enhancement nodes.
A training module 70, configured to train the BLS network to obtain a CFS-GA-BLS fault diagnosis model based on the optimal characteristics and the optimal parameters.
And a diagnosis module 80, configured to input the test set image into the CFS-GA-BLS fault diagnosis model, so as to output a fault infrared diagnosis result of the bearing to be diagnosed.
Further, the conversion module 30 specifically includes:
a gray level unit 31 for graying the picture of interest;
a denoising unit 32, configured to denoise the image of interest after graying by median filtering with a filter having a size of 3×3;
a calculating unit 33, configured to calculate a local mean and a standard deviation according to the brightness distribution of different regions of the image of interest after denoising, and calculate a local threshold based on the local mean and the standard deviation;
a weighting unit 34, configured to obtain an optimal threshold value by using a gaussian weighted average algorithm based on the local threshold value;
A conversion unit 35, configured to convert the picture of interest into a binary image according to the optimal threshold.
Further, the iteration module 50 specifically includes:
a generating unit 51, configured to calculate a correlation coefficient between the second-order statistical features using CFS, and generate a correlation coefficient matrix;
a spreading unit 52, configured to spread the correlation coefficient matrix into a one-dimensional vector by using a stack operation;
a selecting unit 53, configured to select, based on the ranking of the one-dimensional vectors, a target feature with the largest occurrence number among features with the highest correlation;
a ranking unit 54, configured to rank the target feature at the forefront and take it as the optimal feature of the second-order statistical feature.
Further, the genetic module 60 specifically includes:
a setting unit 61, configured to set parameter ranges of N1, N2, and N3, and generate an initialization population;
a selection unit 62 for selecting a reserve dominant individual based on the advanced fitness of the initialized population;
a genetic unit 63 for deriving progeny from a series of genetic manipulations of said dominant individual remaining;
and the screening unit 64 is used for screening the optimal parameters meeting preset targets from the offspring.
The above-described respective modules may be functional modules or program modules, and may be implemented by software or hardware. For modules implemented in hardware, the various modules described above may be located in the same processor; or the above modules may be located in different processors in any combination.
Example 3
The bearing failure infrared diagnostic method described in connection with fig. 1 may be implemented by an electronic device. Fig. 6 is a schematic diagram of the hardware structure of the electronic device according to the present embodiment.
The electronic device may include a processor 91 and a memory 92 storing computer program instructions.
In particular, the processor 91 may comprise a Central Processing Unit (CPU), or an application specific integrated circuit (Application Specific Integrated Circuit, abbreviated as ASIC), or may be configured as one or more integrated circuits embodying the present application.
Memory 92 may include, among other things, mass storage for data or instructions. By way of example, and not limitation, memory 92 may comprise a Hard Disk Drive (HDD), floppy Disk Drive, solid state Drive (Solid State Drive, SSD), flash memory, optical Disk, magneto-optical Disk, tape, or universal serial bus (Universal Serial Bus, USB) Drive, or a combination of two or more of the foregoing. The memory 92 may include removable or non-removable (or fixed) media, where appropriate. The memory 92 may be internal or external to the data processing apparatus, where appropriate. In a particular embodiment, the memory 92 is a Non-Volatile (Non-Volatile) memory. In particular embodiments, memory 92 includes Read-Only Memory (ROM) and random access Memory (Random Access Memory, RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (Programmable Read-Only Memory, abbreviated PROM), an erasable PROM (Erasable Programmable Read-Only Memory, abbreviated EPROM), an electrically erasable PROM (Electrically Erasable Programmable Read-Only Memory, abbreviated EEPROM), an electrically rewritable ROM (Electrically Alterable Read-Only Memory, abbreviated EAROM), or a FLASH Memory (FLASH), or a combination of two or more of these. The RAM may be Static Random-Access Memory (SRAM) or dynamic Random-Access Memory (Dynamic Random Access Memory DRAM), where the DRAM may be a fast page mode dynamic Random-Access Memory (Fast Page Mode Dynamic Random Access Memory FPMDRAM), extended data output dynamic Random-Access Memory (Extended Date Out Dynamic Random Access Memory EDODRAM), synchronous dynamic Random-Access Memory (Synchronous Dynamic Random-Access Memory SDRAM), or the like, as appropriate.
Memory 92 may be used to store or cache various data files that need to be processed and/or communicated, as well as possible computer program instructions for execution by processor 91.
The processor 91 reads and executes the computer program instructions stored in the memory 92 to realize the bearing failure infrared diagnosis method of embodiment 1 described above.
In some of these embodiments, the electronic device may also include a communication interface 93 and a bus 90. As shown in fig. 6, the processor 91, the memory 92, and the communication interface 93 are connected to each other through the bus 90 and perform communication with each other.
The communication interface 93 is used to enable communication between modules, devices, units and/or units in the present application. The communication interface 93 may also enable communication with other components such as: and the external equipment, the image/data acquisition equipment, the database, the external storage, the image/data processing workstation and the like are used for data communication.
Bus 90 includes hardware, software, or both that couple the components of the device to one another. Bus 90 includes, but is not limited to, at least one of: data Bus (Data Bus), address Bus (Address Bus), control Bus (Control Bus), expansion Bus (Expansion Bus), local Bus (Local Bus). By way of example, and not limitation, bus 90 may include a graphics acceleration interface (Accelerated Graphics Port), abbreviated AGP, or other graphics Bus, an enhanced industry standard architecture (Extended Industry Standard Architecture, abbreviated EISA) Bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an industry standard architecture (Industry Standard Architecture, ISA) Bus, a wireless bandwidth (InfiniBand) interconnect, a Low Pin Count (LPC) Bus, a memory Bus, a micro channel architecture (Micro Channel Architecture, abbreviated MCa) Bus, a peripheral component interconnect (Peripheral Component Interconnect, abbreviated PCI) Bus, a PCI-Express (PCI-X) Bus, a serial advanced technology attachment (Serial Advanced Technology Attachment, abbreviated SATA) Bus, a video electronics standards association local (Video Electronics Standards Association Local Bus, abbreviated VLB) Bus, or other suitable Bus, or a combination of two or more of the foregoing. Bus 90 may include one or more buses, where appropriate. Although a particular bus is described and illustrated, the present application contemplates any suitable bus or interconnect.
The electronic device can acquire the bearing failure infrared diagnosis system and execute the bearing failure infrared diagnosis method of the embodiment 1.
In addition, in combination with the bearing failure infrared diagnosis method in the above embodiment 1, the present application can be implemented by providing a storage medium. The storage medium having stored thereon computer program instructions; the computer program instructions, when executed by a processor, implement the bearing failure infrared diagnostic method of embodiment 1 described above.
The technical features of the above-described embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above-described embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing description of the preferred embodiments of the application is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the application.
Claims (9)
1. An infrared bearing fault diagnosis method is characterized by comprising the following steps:
collecting multi-frame infrared images of the running state of a bearing to be diagnosed of a preset type through a rolling bearing fault experiment table; the multi-frame infrared image is divided into a training set image and a testing set image;
Cutting according to a temperature change area in the training set image to obtain an interested picture;
preprocessing the picture of interest to convert the picture of interest into a binary image;
extracting second-order statistical features from the binary image based on a gray level co-occurrence matrix formula; the second-order statistical features comprise energy, contrast, entropy, mean value, variance and correlation;
performing feature iterative optimization on the second-order statistical features to obtain optimal features; specifically, calculating correlation coefficients between the second-order statistical features by adopting CFS, and generating a correlation coefficient matrix; expanding the correlation coefficient matrix into a one-dimensional vector by using stack operation; selecting a target feature with the largest occurrence frequency from the features with the highest correlation based on the sequence of the one-dimensional vectors; sequencing the target features at the forefront and taking the target features as optimal features of the second-order statistical features;
optimizing the BLS network parameters through a genetic algorithm to obtain optimal parameters; the BLS network parameters comprise the number N1 of each window characteristic node, the number N2 of windows of the characteristic nodes and the number N3 of enhancement nodes;
training the BLS network based on the optimal characteristics and the optimal parameters to obtain a CFS-GA-BLS fault diagnosis model;
And inputting the test set image into the CFS-GA-BLS fault diagnosis model so as to output a fault infrared diagnosis result of the bearing to be diagnosed.
2. The method according to claim 1, wherein the predetermined categories of bearing operating conditions to be diagnosed include health, cage failure, inner ring 0.5mm failure, inner ring 1.0mm failure, inner ring 1.5mm failure, outer ring 0.5mm failure, outer ring 1.0mm failure, outer ring 1.5mm failure, rolling element pitting failure.
3. The method for infrared diagnosis of bearing failure according to claim 1, wherein the step of preprocessing the picture of interest to be converted into a binary image specifically comprises:
graying the picture of interest;
carrying out median filtering denoising treatment on the interested picture subjected to graying treatment by adopting a filter with the size of 3 multiplied by 3;
calculating local mean and standard deviation of the image according to brightness distribution of different areas of the image of interest after denoising treatment, and calculating a local threshold value based on the local mean and standard deviation;
obtaining an optimal threshold value by adopting a Gaussian weighted average algorithm based on the local threshold value;
and converting the interested picture into a binary image according to the optimal threshold value.
4. The infrared bearing fault diagnosis method according to claim 1, wherein the gray level co-occurrence matrix formula is specifically as follows:
；
in the method, in the process of the invention,P(i,j,δ,θ) Representing gray values asiAndjis at a distance from each otherδAnd angle ofθIs used to determine the probability of the occurrence of a location of (c),N(i, j,δ,θ) Representing gray values asiAndjis at a distance from each otherδAnd angle ofθIs provided with a number of positions in the pattern,Gthe gray level number of the binary image, i.e., 256.
5. The infrared diagnosis method of bearing faults according to claim 1, wherein each second-order statistical feature is a set of 72 features of total of pixel distance of 1, 2 and 3, angle of 0 ℃, 45 ℃, 90 ℃ and 135 ℃ respectively; said energyf 1 Said contrast ratiof 2 Said entropyf 3 The mean value off 4 Said variance isf 5 The correlation isf 6 The calculation formulas of (a) are respectively as follows:
；/>；/>；
；/>；/>；
in the method, in the process of the invention,N g representing the number of different gray levels in the binary image,p（i, j) Representing the first degree in the normalized gray level co-occurrence matrixi, j) The number of objects to be achieved in the process,nrepresents the number of the row of the codes,μ x 、μ y 、σ x andσ y representation ofp x （i) And (3) withp y （j) Is defined as the mean value and standard deviation of (c),p x （i）、p y （j) Respectively represent the boundary probabilities in the transverse direction and the longitudinal direction,μis thatμ x Andμ y average value of (2).
6. The method for infrared diagnosis of bearing failure according to claim 1, wherein the step of optimizing the BLS network parameter by genetic algorithm to obtain the optimal parameter comprises:
Setting parameter ranges of N1, N2 and N3, and generating an initialization population;
selecting a reserve dominant individual based on the advanced fitness of the initialized population;
subjecting the remaining dominant individual to a series of genetic manipulations to derive progeny;
and screening out optimal parameters meeting preset targets from the offspring.
7. An infrared bearing failure diagnostic system, comprising:
the acquisition module is used for acquiring multi-frame infrared images of the running state of the bearing to be diagnosed of a preset type through the rolling bearing fault experiment table; the multi-frame infrared image is divided into a training set image and a testing set image;
the clipping module is used for clipping according to the temperature change area appearing in the training set image to obtain an interested picture;
the conversion module is used for preprocessing the interested picture and converting the interested picture into a binary image;
the extraction module is used for extracting second-order statistical features from the binary image based on a gray level co-occurrence matrix formula; the second-order statistical features comprise energy, contrast, entropy, mean value, variance and correlation;
the iteration module is used for carrying out characteristic iteration optimization aiming at the second-order statistical characteristics to obtain optimal characteristics; the iteration module specifically comprises:
The generation unit is used for calculating the correlation coefficient between the second-order statistical features by adopting CFS and generating a correlation coefficient matrix;
the expanding unit is used for expanding the correlation coefficient matrix into a one-dimensional vector by using stack operation;
a selecting unit, configured to select a target feature with the largest occurrence number from features with the highest correlation based on the ranking of the one-dimensional vectors;
the sorting unit is used for sorting the target features to the forefront and taking the target features as optimal features of the second-order statistical features;
the genetic module is used for optimizing the BLS network parameters through a genetic algorithm to obtain optimal parameters; the BLS network parameters comprise the number N1 of each window characteristic node, the number N2 of windows of the characteristic nodes and the number N3 of enhancement nodes;
the training module is used for training the BLS network based on the optimal characteristics and the optimal parameters to obtain a CFS-GA-BLS fault diagnosis model;
and the diagnosis module is used for inputting the test set image into the CFS-GA-BLS fault diagnosis model so as to output the fault infrared diagnosis result of the bearing to be diagnosed.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the bearing failure infrared diagnostic method according to any one of claims 1 to 6 when executing the computer program.
9. A storage medium having stored thereon a computer program, which when executed by a processor implements the bearing failure infrared diagnostic method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310677567.1A CN116402825B (en) | 2023-06-09 | 2023-06-09 | Bearing fault infrared diagnosis method, system, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310677567.1A CN116402825B (en) | 2023-06-09 | 2023-06-09 | Bearing fault infrared diagnosis method, system, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116402825A CN116402825A (en) | 2023-07-07 |
| CN116402825B true CN116402825B (en) | 2023-08-11 |
Family
ID=87014670
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310677567.1A Active CN116402825B (en) | 2023-06-09 | 2023-06-09 | Bearing fault infrared diagnosis method, system, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116402825B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117312930A (en) * | 2023-11-29 | 2023-12-29 | 南京邮电大学 | Transformer zero sample fault diagnosis method, device, storage medium and equipment |
| CN117611654B (en) * | 2024-01-24 | 2024-04-16 | 深圳市爱贝科精密工业股份有限公司 | Spindle thermal elongation testing method and testing device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109934206A (en) * | 2019-04-08 | 2019-06-25 | 中国矿业大学(北京) | A kind of rotary machinery fault diagnosis method under non-stationary operating condition |
| WO2021253722A1 (en) * | 2020-06-19 | 2021-12-23 | 中国科学院深圳先进技术研究院 | Medical image reconstruction technology method and apparatus, storage medium and electronic device |
| CN115293206A (en) * | 2022-08-09 | 2022-11-04 | 贵州大学 | Bearing fault diagnosis method based on sparrow algorithm optimization width learning system |
| CN116051911A (en) * | 2023-03-29 | 2023-05-02 | 北京大学 | Small sample bearing vibration image data fault diagnosis method based on uncertainty learning |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110426415A (en) * | 2019-07-15 | 2019-11-08 | 武汉大学 | Based on thermal fault detection method inside depth convolutional neural networks and the oil-immersed transformer of image segmentation |
-
2023
- 2023-06-09 CN CN202310677567.1A patent/CN116402825B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109934206A (en) * | 2019-04-08 | 2019-06-25 | 中国矿业大学(北京) | A kind of rotary machinery fault diagnosis method under non-stationary operating condition |
| WO2021253722A1 (en) * | 2020-06-19 | 2021-12-23 | 中国科学院深圳先进技术研究院 | Medical image reconstruction technology method and apparatus, storage medium and electronic device |
| CN115293206A (en) * | 2022-08-09 | 2022-11-04 | 贵州大学 | Bearing fault diagnosis method based on sparrow algorithm optimization width learning system |
| CN116051911A (en) * | 2023-03-29 | 2023-05-02 | 北京大学 | Small sample bearing vibration image data fault diagnosis method based on uncertainty learning |
Non-Patent Citations (1)
| Title |
|---|
| 基于GLCM-SDAE的滚动轴承故障诊断方法;须颖;李昊东;安冬;;沈阳建筑大学学报(自然科学版)(第04期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116402825A (en) | 2023-07-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116402825B (en) | Bearing fault infrared diagnosis method, system, electronic equipment and storage medium | |
| CN112435221A (en) | Image anomaly detection method based on generative confrontation network model | |
| CN110008948B (en) | Hyperspectral image target detection method based on variational self-coding network | |
| CN111127387B (en) | Quality evaluation method for reference-free image | |
| CN105260998A (en) | MCMC sampling and threshold low-rank approximation-based image de-noising method | |
| CN109726649B (en) | Remote sensing image cloud detection method and system and electronic equipment | |
| CN108830289B (en) | Image clustering method and device based on improved fuzzy C-means clustering | |
| CN115661649B (en) | BP neural network-based shipborne microwave radar image oil spill detection method and system | |
| CN114169377A (en) | G-MSCNN-based fault diagnosis method for rolling bearing in noisy environment | |
| CN116823725A (en) | Aeroengine blade surface defect detection method based on support vector machine | |
| CN113378620B (en) | Cross-camera pedestrian re-identification method in surveillance video noise environment | |
| CN112884721B (en) | Abnormality detection method, abnormality detection system and computer-readable storage medium | |
| CN108830829B (en) | Non-reference quality evaluation algorithm combining multiple edge detection operators | |
| CN106934398A (en) | Image de-noising method based on super-pixel cluster and rarefaction representation | |
| CN114037001A (en) | Mechanical pump small sample fault diagnosis method based on WGAN-GP-C and metric learning | |
| CN111428772B (en) | Photovoltaic system depth anomaly detection method based on k-nearest neighbor adaptive voting | |
| Ji et al. | An automatic bad band pre-removal method for hyperspectral imagery | |
| CN116229254A (en) | Remote sensing extraction method for offshore buoyant raft and deep water culture area | |
| CN115791174A (en) | Rolling bearing abnormity diagnosis method and system, electronic equipment and storage medium | |
| Zhang et al. | Steel strip surface inspection through the combination of feature selection and multiclass classifiers | |
| CN111899226B (en) | Hyperspectral image target prior optimization method based on multitask sparse learning | |
| CN111797732B (en) | Video motion identification anti-attack method insensitive to sampling | |
| CN114399635A (en) | Image two-classification ensemble learning method based on feature definition and deep learning | |
| CN111652246A (en) | Image self-adaptive sparsization representation method and device based on deep learning | |
| CN117095208B (en) | Lightweight scene classification method for photoelectric pod reconnaissance image |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |