Prediction of Fusion Hole Perforation Based on Arc Characteristics of Front Image in Backing Welding
In one-side welding with back-formation, the weld is penetrated after the fusion hole is perforated. Therefore, judging whether the fusion hole is perforated is very important to realize autocontrol of penetration in one-side welding with back-formation process. Previous researches mainly focused on...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-10-01
|
| Series: | Materials |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1996-1944/13/21/4706 |
| _version_ | 1797550157402734592 |
|---|---|
| author | Yu Cao Xiaofei Wang Xu Yan Chuanbao Jia Jinqiang Gao |
| author_facet | Yu Cao Xiaofei Wang Xu Yan Chuanbao Jia Jinqiang Gao |
| author_sort | Yu Cao |
| collection | DOAJ |
| description | In one-side welding with back-formation, the weld is penetrated after the fusion hole is perforated. Therefore, judging whether the fusion hole is perforated is very important to realize autocontrol of penetration in one-side welding with back-formation process. Previous researches mainly focused on the morphological characteristics of the molten pool and fusion hole to judge the weld penetration state. Sometimes it is difficult to obtain the morphological characteristics of the molten pool, keyhole and fusion hole and image processing is complex. In this paper, a visual detection system of fusion holes based on visual sensing is constructed to obtain the real-time fusion hole images in backing welding. It is found that the arc characteristics in the front images contain abundant information about the perforation of fusion hole. An image processing program is developed based on MATLAB software, and the arc characteristic parameters in front images are obtained. Taking the arc characteristic parameters as the input, obtaining the penalty function and the kernel function parameters through the cross validation and grid search method, a prediction model of fusion hole perforation based on the support vector machine is put forward. The accuracy for prediction samples is 88%. By analyzing the misidentified samples, it is found that some of the newly perforated images are predicted as nonperforated ones, which has little influence on the penetration control of the weld. |
| first_indexed | 2024-03-10T15:25:32Z |
| format | Article |
| id | doaj.art-51acfd22dccd43fcb1382e512293f071 |
| institution | Directory Open Access Journal |
| issn | 1996-1944 |
| language | English |
| last_indexed | 2024-03-10T15:25:32Z |
| publishDate | 2020-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Materials |
| spelling | doaj.art-51acfd22dccd43fcb1382e512293f0712023-11-20T18:04:46ZengMDPI AGMaterials1996-19442020-10-011321470610.3390/ma13214706Prediction of Fusion Hole Perforation Based on Arc Characteristics of Front Image in Backing WeldingYu Cao0Xiaofei Wang1Xu Yan2Chuanbao Jia3Jinqiang Gao4MOE Key Lab for Liquid–Solid Structure Evolution and Materials Processing, Shandong University, Jinan 250061, ChinaMOE Key Lab for Liquid–Solid Structure Evolution and Materials Processing, Shandong University, Jinan 250061, ChinaMOE Key Lab for Liquid–Solid Structure Evolution and Materials Processing, Shandong University, Jinan 250061, ChinaMOE Key Lab for Liquid–Solid Structure Evolution and Materials Processing, Shandong University, Jinan 250061, ChinaInstitute of Materials Joining, Shandong University, Jinan 250061, ChinaIn one-side welding with back-formation, the weld is penetrated after the fusion hole is perforated. Therefore, judging whether the fusion hole is perforated is very important to realize autocontrol of penetration in one-side welding with back-formation process. Previous researches mainly focused on the morphological characteristics of the molten pool and fusion hole to judge the weld penetration state. Sometimes it is difficult to obtain the morphological characteristics of the molten pool, keyhole and fusion hole and image processing is complex. In this paper, a visual detection system of fusion holes based on visual sensing is constructed to obtain the real-time fusion hole images in backing welding. It is found that the arc characteristics in the front images contain abundant information about the perforation of fusion hole. An image processing program is developed based on MATLAB software, and the arc characteristic parameters in front images are obtained. Taking the arc characteristic parameters as the input, obtaining the penalty function and the kernel function parameters through the cross validation and grid search method, a prediction model of fusion hole perforation based on the support vector machine is put forward. The accuracy for prediction samples is 88%. By analyzing the misidentified samples, it is found that some of the newly perforated images are predicted as nonperforated ones, which has little influence on the penetration control of the weld.https://www.mdpi.com/1996-1944/13/21/4706one-side welding with back-formationfusion holevisual sensingpredictionarc characteristics |
| spellingShingle | Yu Cao Xiaofei Wang Xu Yan Chuanbao Jia Jinqiang Gao Prediction of Fusion Hole Perforation Based on Arc Characteristics of Front Image in Backing Welding Materials one-side welding with back-formation fusion hole visual sensing prediction arc characteristics |
| title | Prediction of Fusion Hole Perforation Based on Arc Characteristics of Front Image in Backing Welding |
| title_full | Prediction of Fusion Hole Perforation Based on Arc Characteristics of Front Image in Backing Welding |
| title_fullStr | Prediction of Fusion Hole Perforation Based on Arc Characteristics of Front Image in Backing Welding |
| title_full_unstemmed | Prediction of Fusion Hole Perforation Based on Arc Characteristics of Front Image in Backing Welding |
| title_short | Prediction of Fusion Hole Perforation Based on Arc Characteristics of Front Image in Backing Welding |
| title_sort | prediction of fusion hole perforation based on arc characteristics of front image in backing welding |
| topic | one-side welding with back-formation fusion hole visual sensing prediction arc characteristics |
| url | https://www.mdpi.com/1996-1944/13/21/4706 |
| work_keys_str_mv | AT yucao predictionoffusionholeperforationbasedonarccharacteristicsoffrontimageinbackingwelding AT xiaofeiwang predictionoffusionholeperforationbasedonarccharacteristicsoffrontimageinbackingwelding AT xuyan predictionoffusionholeperforationbasedonarccharacteristicsoffrontimageinbackingwelding AT chuanbaojia predictionoffusionholeperforationbasedonarccharacteristicsoffrontimageinbackingwelding AT jinqianggao predictionoffusionholeperforationbasedonarccharacteristicsoffrontimageinbackingwelding |