%0 Journal Article
%T 漆层下钢材腐蚀红外图像特征提取算法对比研究<br>Comparison Research on Feature Extraction Algorithms for Infrared Images for Paint Coated Steel Corrosion
%A 张艳芳
%A 姚盛清
%A 薛超
%A 肖智中
%A 丁松
%J Applied Physics
%P 569-583
%@ 2160-7575
%D 2022
%I Hans Publishing
%R 10.12677/APP.2022.1211067
%X 防腐漆层下材料腐蚀是对钢材性能和钢结构健康的一大挑战。本文对基于涡流脉冲热成像技术的涂层下钢材腐蚀无损检测与评估技术展开研究，提出温度偏度提取方法，并将其与主成分分析、相位分析和峰度特征提取等方法进行对比，用于漆层下钢材腐蚀的表征和评估。对于早期腐蚀，平均偏度显示出比第二主成分及相位特征更好的灵敏度；然而，归一化的第二主成分具有更大的检测范围，可以识别0~10个月的腐蚀。腐蚀区域温度变化分析表明，电导率对早期腐蚀感应加热和热分布起主导作用。但是长期腐蚀(10个月以上)对焦耳热的分布和耗散影响机制复杂，上述方法均不能很好地识别和评估腐蚀状态。<br />
Paint coated steel corrosion is a big challenge to steel performance and structure health. This paper investigates the nondestructive testing and evaluation based on eddy current pulsed thermography (ECPT) technique for paint coated steel corrosion. Temperature skewness is proposed and compared with principal component analysis (PCA), phase analysis, and kurtosis feature extraction methods, which are applied for paint coated corrosion characterization and evaluation. For early stage corrosion, the averaged skewness shows better sensitivity than the 2nd principal component (2nd PC) and phase analysis performance. However, the normalized 2nd PC presents good sensitivity and the best measurement scale of 0 to 10 months of corrosion. The temperature variation analysis indicates that the electrical conductivity dominates the induced heating and heat distribution for early stage corrosion. For long-time corrosion, more than 10 months, the mechanism of the influence on Joule heat distribution and emission is so complex that the proposed approaches can not identify and evaluate the corrosion state.
%K 漆层下腐蚀，无损检测，脉冲涡流热成像，特征提取<br>Paint Coated Corrosion
%K Nondestructive Test
%K Eddy Current Pulsed Thermography
%K Feature Extraction
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=57998