1. Single insulator infrared image fault diagnosis method based on sparse presentation
- Author
-
Liu Yunpeng, Hu Chengbo, Xu Changfu, Pei Shaotong, Ji Xinxin, and Xu Jiayuan
- Subjects
Condensed Matter::Quantum Gases ,Infrared image ,Engineering ,Noise immunity ,Infrared ,business.industry ,Pattern recognition ,Insulator (electricity) ,Sparse approximation ,Homomorphic filtering ,Computer Science::Computer Vision and Pattern Recognition ,Diagnostic model ,Electronic engineering ,Condensed Matter::Strongly Correlated Electrons ,Artificial intelligence ,Single image ,business - Abstract
In this paper, taking typical porcelain insulators for research objects, the deteriorated diagnosis model of single insulator infrared image is studied. By infrared imaging, layered cutting, homomorphic filtering de-noising, a set of single insulator to be processed is obtained. Based on the sparse characteristics of single insulator infrared image, building an over-complete dictionary with the single image as the constituent atom. The sparse representation coefficient of the identified image in the dictionary is calculated, diagnosing identified single insulator according to the nonzero term in the coefficient. Through test and analysis, the designed diagnostic model of single insulator has high classification accuracy. Furthermore, it also has stronger noise immunity to the background and a better recognition ability for the initial heating of the deteriorated insulator.
- Published
- 2017