Using Machine Learning for Aerostructure Surface Damage Digital Reconstruction
Aerostructure surface damage inspection is carried out over the whole life-cycle using legacy processes and recording during maintenance. The inspection techniques record the detailed history of the damage and repair. However it remains elusive to predict the location of future damage on the aerostr...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-01-01
|
| Series: | Aerospace |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2226-4310/12/1/72 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832589506354610176 |
|---|---|
| author | Yijia Wu Hon Ping Tang Anthony Mannion Robert Voyle Ying Xin |
| author_facet | Yijia Wu Hon Ping Tang Anthony Mannion Robert Voyle Ying Xin |
| author_sort | Yijia Wu |
| collection | DOAJ |
| description | Aerostructure surface damage inspection is carried out over the whole life-cycle using legacy processes and recording during maintenance. The inspection techniques record the detailed history of the damage and repair. However it remains elusive to predict the location of future damage on the aerostructure surface. In this paper, we work up a novel simulation technique based on the results of machine learning analysis for prediction of the reference location of future aerostructure surface damage. First, we use the support vector machine (SVM) and k-nearest neighbor (KNN) to analyze the damage on three B777-200 aircraft and find that the classification accuracy can range from 75.1% to 86%. Then, we use the prediction result of a feedforward neural network (FNN) to simulate the damage structure and the mapping relationship to explore its reconstructive possibility. We show that the aerostructure surface damage can be reconstructed by machine learning. Moreover, the aerostructure surface damage heat map obtained by reconstruction can be prepared for further image recognition analysis in the future. |
| format | Article |
| id | doaj-art-dc78826822cb4d31be14d11334edfb7d |
| institution | Kabale University |
| issn | 2226-4310 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Aerospace |
| spelling | doaj-art-dc78826822cb4d31be14d11334edfb7d2025-01-24T13:15:44ZengMDPI AGAerospace2226-43102025-01-011217210.3390/aerospace12010072Using Machine Learning for Aerostructure Surface Damage Digital ReconstructionYijia Wu0Hon Ping Tang1Anthony Mannion2Robert Voyle3Ying Xin4Aviation Services Research Centre, Room 241, Block X, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, ChinaAviation Services Research Centre, Room 241, Block X, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, ChinaAviation Services Research Centre, Room 241, Block X, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, ChinaAviation Services Research Centre, Room 241, Block X, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, ChinaAviation Services Research Centre, Room 241, Block X, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, ChinaAerostructure surface damage inspection is carried out over the whole life-cycle using legacy processes and recording during maintenance. The inspection techniques record the detailed history of the damage and repair. However it remains elusive to predict the location of future damage on the aerostructure surface. In this paper, we work up a novel simulation technique based on the results of machine learning analysis for prediction of the reference location of future aerostructure surface damage. First, we use the support vector machine (SVM) and k-nearest neighbor (KNN) to analyze the damage on three B777-200 aircraft and find that the classification accuracy can range from 75.1% to 86%. Then, we use the prediction result of a feedforward neural network (FNN) to simulate the damage structure and the mapping relationship to explore its reconstructive possibility. We show that the aerostructure surface damage can be reconstructed by machine learning. Moreover, the aerostructure surface damage heat map obtained by reconstruction can be prepared for further image recognition analysis in the future.https://www.mdpi.com/2226-4310/12/1/72machine learningaerostructuredamagesurfacemaintenancelife-cycle |
| spellingShingle | Yijia Wu Hon Ping Tang Anthony Mannion Robert Voyle Ying Xin Using Machine Learning for Aerostructure Surface Damage Digital Reconstruction Aerospace machine learning aerostructure damage surface maintenance life-cycle |
| title | Using Machine Learning for Aerostructure Surface Damage Digital Reconstruction |
| title_full | Using Machine Learning for Aerostructure Surface Damage Digital Reconstruction |
| title_fullStr | Using Machine Learning for Aerostructure Surface Damage Digital Reconstruction |
| title_full_unstemmed | Using Machine Learning for Aerostructure Surface Damage Digital Reconstruction |
| title_short | Using Machine Learning for Aerostructure Surface Damage Digital Reconstruction |
| title_sort | using machine learning for aerostructure surface damage digital reconstruction |
| topic | machine learning aerostructure damage surface maintenance life-cycle |
| url | https://www.mdpi.com/2226-4310/12/1/72 |
| work_keys_str_mv | AT yijiawu usingmachinelearningforaerostructuresurfacedamagedigitalreconstruction AT honpingtang usingmachinelearningforaerostructuresurfacedamagedigitalreconstruction AT anthonymannion usingmachinelearningforaerostructuresurfacedamagedigitalreconstruction AT robertvoyle usingmachinelearningforaerostructuresurfacedamagedigitalreconstruction AT yingxin usingmachinelearningforaerostructuresurfacedamagedigitalreconstruction |