CN115656335A - Method for quickly detecting and identifying internal defects of bearing ring - Google Patents
Method for quickly detecting and identifying internal defects of bearing ring Download PDFInfo
- Publication number
- CN115656335A CN115656335A CN202211233171.XA CN202211233171A CN115656335A CN 115656335 A CN115656335 A CN 115656335A CN 202211233171 A CN202211233171 A CN 202211233171A CN 115656335 A CN115656335 A CN 115656335A
- Authority
- CN
- China
- Prior art keywords
- bearing ring
- defect
- scanning
- characteristic information
- ultrasonic
- 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
Landscapes
- Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)
Abstract
The invention relates to a method for quickly detecting and identifying internal defects of a bearing ring. The bearing ring to be detected is immersed in water, the focusing ultrasonic probe is fixed above the bearing ring to be detected through the clamping and moving device, the ultrasonic probe emits ultrasonic waves and receives reflected signals, the signals are transmitted into a computer, and obtained ultrasonic data are used for A, B, C scanning imaging. Extracting defect echo envelopes from the scanning data A through complex wavelet transformation, and obtaining characteristic information such as envelope slope, area, amplitude and the like from the defect echo envelopes; preprocessing B, C scanning data through graying, threshold segmentation and the like, and obtaining relevant characteristic information such as length, invariant moment, circularity, rectangularity and the like from a binary image to reflect defect depth information and morphology characteristics. The defect types can be divided into three types, namely holes, cracks and slag inclusion, the characteristic information of all known samples is divided into a training set and a testing set, and a Bayesian classifier is constructed in a cross-training mode. The classifier is applied to defect identification of an unknown bearing ring sample, water immersion ultrasound A, B, C scanning is carried out on the unknown bearing ring sample, characteristic information is collected, and whether a defect exists in the bearing ring sample and the damage type of the defect when the defect exists are rapidly judged through calculation of the Bayesian classifier.
Description
Technical Field
The invention relates to the field of nondestructive testing, in particular to a method for quickly detecting and identifying internal defects of a bearing ring.
Background
The bearing is an important basic part in modern industrial devices, is widely applied to industries such as automobiles, motors, spaceflight, machinery and the like, and in order to ensure that the performance of the bearing reaches the standard and avoid serious consequences caused by bearing failure, the defect detection of a bearing ring is required to ensure the product quality.
In the prior art, the defect detection and identification of the bearing ring are realized by detecting internal defect damage through eddy current, magnetic powder and ultrasonic detection. Eddy current and magnetic particle detection can only measure near-surface defects and require the detected workpiece to be conductive, so that the limitation is large, and the water immersion ultrasonic detection technology can be widely applied to the industrial field of bearing flaw detection because non-contact nondestructive detection can be realized. The water immersion ultrasonic detection is to transmit ultrasonic waves through a water immersion focusing ultrasonic probe, enter a bearing ring through a water layer and collect returned ultrasonic information to perform data analysis so as to achieve the purpose of defect detection.
With the continuous development of modern industry, the role of a bearing as an important part is increasing day by day, the processing and production of the bearing has a mature industrial chain, but the detection and identification of the defects of the bearing still do not form a system with higher automation level, and the bearing still stays at the stage of manual detection at present, and the mechanization level is low. The principle of the immersion ultrasonic detection is that ultrasonic waves can be reflected when encountering defects, the defects can be positioned and measured through received echo waveforms, and the judgment of the types of the internal defects of the bearing ring through the defect echo waveforms is usually based on the experience judgment of technical personnel, so that the types of the internal defects of the bearing ring are difficult to accurately quantify, the manual interference is large, and the immersion ultrasonic detection is not beneficial to the automation process of the bearing industry.
Disclosure of Invention
The invention provides a method for rapidly detecting and identifying internal defects of a bearing ring, aiming at the problems that the method for judging the types of the internal defects of the bearing ring through water immersion ultrasonic detection has low automation level and is difficult to get rid of interference of human factors, and aiming at reducing the interference of the human factors in the defect identification process through a machine learning mode and improving the efficiency and the automation level of a quality detection link of the bearing ring.
The invention effectively realizes the defect detection and classification identification of the inside of the bearing ring by a water immersion ultrasonic detection mode based on the Bayesian classification principle. The bearing ring to be detected is immersed in water, the focused ultrasonic probe is fixed above the bearing ring to be detected through the clamping and moving device, the ultrasonic probe emits ultrasonic waves and receives reflected signals to transmit the signals to a computer, and obtained ultrasonic data are used for A, B, C scanning imaging. Extracting defect echo envelopes from the scanning data A through complex wavelet transformation, and obtaining characteristic information such as envelope slope, area, amplitude and the like from the defect echo envelopes; preprocessing B, C scanning data through gray threshold segmentation and the like, and obtaining related characteristic information such as length, invariant moment, circularity, rectangularity and the like from an image to reflect defect depth information and morphology characteristics. The defect types can be divided into three types, namely holes, cracks and slag inclusion, the characteristic information of all known samples is divided into a training set and a testing set, and a Bayesian classifier is constructed in a cross-training mode. The classifier is applied to defect identification of an unknown bearing ring sample, water immersion ultrasound A, B, C scanning is carried out on the unknown bearing ring sample, characteristic information is collected, and whether a defect exists in the bearing ring sample and the damage type of the defect when the defect exists are rapidly judged through calculation of the Bayesian classifier.
The invention has the beneficial effects that:
1. the Bayes classifier is the simplest and most effective classification statistical model under small sample data, a classifier model is constructed through learning of training samples, and the classifier is used for classifying unknown samples, so that the recognition accuracy is high, the operation speed is high, and the problem that the known samples of the defects of the bearing ring are insufficient is not required to be considered too much.
2. The A, B, C scanning of water immersion ultrasonic detection is carried out on the bearing ring sample, the characteristic information of the sample defect can be obtained more comprehensively, and compared with a single scanning mode, the accuracy of the classifier can be improved to a great extent.
3. So far, the detection and identification of the internal defects of the bearing ring mainly depend on manpower, and the method can independently realize high-efficiency detection and identification of the internal defects of the bearing ring, reduce the interference of human factors, reduce the labor cost, improve the detection efficiency and generate great economic benefits for society and industry.
4. The method can be used for identifying the internal defects of the bearing ring taking water immersion ultrasonic scanning data as an object, and can also be expanded to other detection fields.
Drawings
FIG. 1 is a schematic diagram of an experiment in the present invention.
The ultrasonic testing device comprises (1) a water immersion focusing ultrasonic probe (2), a bearing ring sample to be tested (3), a water layer (4), a clamping and moving device (5), a water tank (6) and a computer.
FIG. 2 is a flowchart of a Bayesian classifier algorithm of the present invention.
Detailed Description
The invention will be further described with reference to the accompanying drawings.
As shown in fig. 1, a water immersion ultrasonic detection device for a bearing ring comprises (1) a water immersion focusing ultrasonic probe (2), a bearing ring sample to be detected (3), a water layer (4), a clamping and moving device (5), and a water tank (6) computer.
According to the invention, a bearing ring (2) to be detected is placed in a water tank (5) and is immersed in a water layer (3), a focused ultrasonic probe (1) is placed above a bearing through a clamping and moving device (4), the probe transmits ultrasonic waves and receives reflected signals to transmit into a computer (6), and obtained ultrasonic data is used for A, B, C scanning imaging.
The defect recognition principle of the invention is as follows:
when the incident ultrasonic waves encounter defects in the ferrule, a part of the ultrasonic waves return in the original path at the defect to form a defect echo, the ultrasonic signals received and returned by the ultrasonic probe are transmitted into a computer and used for A, B, C scanning imaging, and when the types of the defects encountered by the ultrasonic waves are different, the scanning imaging effects are different, wherein the types of the defects can be divided into three types, namely holes, cracks and slag inclusion.
The invention extracts characteristic parameters such as envelope amplitude, slope, time, image contour, invariant moment and the like from ultrasonic scanning imaging of internal defects of each bearing ring, realizes dimension reduction through a principal component analysis method, reduces redundancy and overlapping of characteristic description, and constructs a Bayes classifier for the characteristic information in a cross training mode: calculating prior probabilities of various defect types, and taking the ratio of the number of the defect types to the total number of samples as a calculation mode of the prior probabilities; calculating the conditional probability of each characteristic parameter, and if the parameters meet Gaussian distribution, integrating the probability density to obtain the conditional probability; and calculating the posterior probability of each type of the sample according to the prior probability and the conditional probability of each characteristic parameter, wherein the maximum probability is the damage type of the sample defect judged by the Bayes classifier, and finally, performing performance test on the classifier by using the unknown bearing ring sample to realize the detection and identification of the unknown sample defect.
The specific process of the Bayesian classification algorithm with the water immersion ultrasonic scanning image as the object is shown in FIG. 2:
1. performing water immersion ultrasonic detection A, B, C scanning imaging on each bearing ring sample, wherein an image A1 is obtained by A scanning, an image B1 is obtained by B scanning, and an image C1 is obtained by C scanning;
2. performing complex wavelet transform on the image A1 to extract a defect echo envelope A2, and performing graying and threshold segmentation on the images B1 and C1 to convert the images into binary images B2 and C2 respectively;
3. extracting characteristic parameters such as start time, end time, maximum amplitude time, slope from start time to maximum amplitude time, slope from maximum amplitude time to end time, area and the like from the defect echo signal envelope A2, and extracting characteristic parameters such as area, perimeter, circularity, rectangularity, invariant moment, hawood diameter, circumscribed rectangle length-width ratio, ratio of circumscribed rectangle short axis to object area, ratio of object area to circumscribed rectangle area and the like from the binary pictures B2 and C2 respectively;
4. performing dimensionality reduction on the N groups of characteristic parameters by using a Principal Component Analysis (PCA) method, and selecting M main characteristic parameters from the N characteristic parameters according to variance contribution rate;
5. dividing all samples into a training set and a testing set, and constructing a Bayesian classifier in a cross training mode;
6. and carrying out performance test on the classifier by using the unknown bearing ring sample to realize the defect detection and identification of the unknown bearing ring.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and all simple modifications, changes and equivalent structural changes made to the above embodiment according to the technical spirit of the present invention still fall within the protection scope of the technical solution of the present invention.
Claims (2)
1. A method for quickly detecting and identifying internal defects of a bearing ring is characterized by comprising the following steps: the bearing ring to be detected is immersed in water, the focused ultrasonic probe is fixed above the bearing ring to be detected through the clamping and moving device, the ultrasonic probe emits ultrasonic waves and receives reflected signals to transmit the signals into a computer, and obtained ultrasonic data are used for A, B, C scanning imaging. Extracting defect echo envelopes from the scanning data A through complex wavelet transformation, and obtaining characteristic information such as envelope slope, area, amplitude and the like from the defect echo envelopes; preprocessing B, C scanning data through graying, threshold segmentation and the like, and obtaining relevant characteristic information such as length, invariant moment, circularity, rectangularity and the like from a binary image to reflect defect depth information and morphology characteristics. The defect types can be divided into three types, namely holes, cracks and slag inclusion, the characteristic information of all known samples is divided into a training set and a testing set, and a Bayesian classifier is constructed in a cross-training mode. The classifier is applied to defect identification of an unknown bearing ring sample, water immersion ultrasonic A, B, C scanning is carried out on the unknown bearing ring sample, characteristic information is collected, and whether a defect exists in the bearing ring sample and the damage type of the defect when the defect exists are quickly judged through calculation of the Bayes classifier.
2. The method for detecting and identifying the internal defect of the bearing ring according to claim 1, wherein: the specific mode of the Bayes classification algorithm with the water immersion ultrasonic scanning image as the object is as follows:
step 1, performing water immersion ultrasonic detection A, B, C scanning imaging on each bearing ring sample, wherein A scanning obtains an image A1, B scanning obtains an image B1, and C scanning obtains an image C1;
step 2, performing complex wavelet transformation on the image A1 to extract a defect echo envelope A2, and performing graying and threshold segmentation on the images B1 and C1 to convert the images into binary images B2 and C2 respectively;
step 3, extracting characteristic parameters such as starting time, ending time, maximum amplitude time, slope from the starting time to the maximum amplitude time, slope from the maximum amplitude time to the ending time, area and the like from the defect echo signal envelope A2, and extracting characteristic parameters such as area, perimeter, circularity, rectangularity, invariant moment, hawood diameter, circumscribed rectangle length-width ratio, ratio of circumscribed rectangle short axis to object area, ratio of object area to circumscribed rectangle area and the like from the binary pictures B2 and C2 respectively;
step 4, performing dimensionality reduction on the N groups of characteristic parameters by using a Principal Component Analysis (PCA) method, and selecting M main characteristic parameters from the N characteristic parameters according to variance contribution rate;
step 5, dividing all samples into a training set and a testing set, and constructing a Bayes classifier in a cross-training mode;
and 6, carrying out performance test on the classifier by using the unknown bearing ring sample to realize the defect detection and identification of the unknown bearing ring.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211233171.XA CN115656335A (en) | 2022-10-10 | 2022-10-10 | Method for quickly detecting and identifying internal defects of bearing ring |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211233171.XA CN115656335A (en) | 2022-10-10 | 2022-10-10 | Method for quickly detecting and identifying internal defects of bearing ring |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115656335A true CN115656335A (en) | 2023-01-31 |
Family
ID=84987924
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211233171.XA Pending CN115656335A (en) | 2022-10-10 | 2022-10-10 | Method for quickly detecting and identifying internal defects of bearing ring |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115656335A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116136393A (en) * | 2023-03-02 | 2023-05-19 | 宁波川原精工机械有限公司 | Bearing ring inner ring detection system and method |
| KR102658875B1 (en) * | 2023-12-13 | 2024-04-18 | 주식회사 딥아이 | Eddy current test data labeling system |
-
2022
- 2022-10-10 CN CN202211233171.XA patent/CN115656335A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116136393A (en) * | 2023-03-02 | 2023-05-19 | 宁波川原精工机械有限公司 | Bearing ring inner ring detection system and method |
| KR102658875B1 (en) * | 2023-12-13 | 2024-04-18 | 주식회사 딥아이 | Eddy current test data labeling system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115656335A (en) | Method for quickly detecting and identifying internal defects of bearing ring | |
| CN110363746B (en) | Ultrasonic nondestructive detection signal classification method based on convolutional neural network | |
| CN111024728B (en) | Railway detection method and system based on computer vision and ultrasonic flaw detection | |
| CN102928435A (en) | Aircraft skin damage identification method and device based on image and ultrasound information fusion | |
| Posilović et al. | Flaw detection from ultrasonic images using YOLO and SSD | |
| CN112215803B (en) | Aluminum plate eddy current inspection image defect segmentation method based on improved generation countermeasure network | |
| CN110231395B (en) | Method and system for identifying broken wire damage of steel wire rope | |
| Zhuang et al. | The automatic rail surface multi-flaw identification based on a deep learning powered framework | |
| Yuan et al. | Automatic recognition and positioning of wheel defects in ultrasonic B-Scan image using artificial neural network and image processing | |
| CN111539489A (en) | Millimeter wave image target detection and identification method | |
| CN107727749A (en) | A kind of ultrasonic quantitative detection method based on wavelet packet fusion feature extraction algorithm | |
| Lei et al. | Ultrasonic pig for submarine oil pipeline corrosion inspection | |
| CN106404893A (en) | Automatic axial magnetic flux leakage defect signal detection method | |
| Chen et al. | A cascaded deep learning approach for detecting pipeline defects via pretrained YOLOv5 and ViT models based on MFL data | |
| CN116773654A (en) | Intelligent pipeline defect detection device and detection method based on circumferential guided waves | |
| CN117173461A (en) | Multi-visual task filling container defect detection method, system and medium | |
| CN112067697A (en) | Laser ultrasonic B scanning-based defect positioning method | |
| Yang et al. | Weld Defect Cascaded Detection Model Based on Bidirectional Multi-scale Feature Fusion and Shape Pre-classification | |
| CN115223044A (en) | End-to-end three-dimensional ground penetrating radar target identification method and system based on deep learning | |
| Medak et al. | Detection of Defective Bolts from Rotational Ultrasonic Scans Using Convolutional Neural Networks | |
| Daniel et al. | Tsallis entropy segmentation and shape feature-based classification of defects in the simulated magnetic flux leakage images of steam generator tubes | |
| CN112508862A (en) | Method for enhancing magneto-optical image of crack by improving GAN | |
| CN103926321A (en) | Method for marking and positioning erosion defects in inner cavity of engine | |
| Shi et al. | CNN based defect recognition model for phased array ultrasonic testing images of electrofusion joints | |
| CN116645371B (en) | Rail surface defect detection method and system based on feature search |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |