Contamination grades detection method of ceramic insulator based on hyperspectral technology

Abstract

Traditional indicators such as equivalent salt deposit density, surface pollution layer conductance and leakage current are mainly used to measure the contamination grades of ceramic insulators, and the detection process is time-consuming and inefficient. Hyperspectral technology can obtain rich information of target spectrum non-contact, and has great application potential in insulator pollution detection. Therefore, a method of ceramic insulator pollution level detection based on hyperspectral technology is proposed in this paper. Firstly, there is a reflection phenomenon when collecting hyperspectral image due to the material of ceramic insulator, so histogram equalization is used to process the hyperspectral image to remove the reflective interference. Then, the hyperspectral image is preprocessed to remove the noise interference. And the successive projections algorithm (SPA) is used to extract the feature of the sample spectral line for removing the redundant information. Finally, the classification model of support vector machine (SVM) is established according to the characteristic spectral lines, and the classification accuracy of ceramic samples is 95%.

Keywords

• ceramic insulator
• hyperspectral technology
• histogram equalization
• successive projections algorithm (spa)
• contamination grades detection
• support vector machine (svm)