Driven by machine learning to intelligent damage recognition of terminal optical components

• Published in: Neural computing & applications Vol. 33; no. 2; pp. 789 - 804
• Main Author: Yin, Xiangbao
• Format: Journal Article
• Language: English
• Published: London Springer London 01.01.2021; Springer Nature B.V
• Subjects: Artificial Intelligence; Classification; Computational Biology/Bioinformatics; Computational Science and Engineering; Computer Science; Damage detection; Data Mining and Knowledge Discovery; Image Processing and Computer Vision; Infrared spectra; Machine learning; Near infrared radiation; Nondestructive testing; Optical components; Preprocessing; Probability and Statistics in Computer Science; S.I. : DPTA Conference 2019; Target detection; Technical services; Damage identification; Infrared nondestructive testing; Terminal optics; Machine learning
• ISSN: 0941-0643; 1433-3058
• DOI: 10.1007/s00521-020-05051-x

In order to realize the terminal optical element online detection system in the Shenguang III system, each optical element in each terminal optical component in the target room is detected. The research on the optical damage of terminal optical components focuses on the search for damage points, the extraction of damage information, and the classification of damage types. In addition, damage classification and identification of terminal optical components are performed through machine learning, and infrared nondestructive testing is used as technical support to improve the identification model and reduce the complexity of the spectral model. After studying the preprocessing and dimensionality reduction methods of near-infrared spectroscopy, this paper compares the effects of different preprocessing methods and screening feature methods and combines different modeling methods to conduct experiments. The research results show that the method proposed in this paper has certain effects.