传感技术学报2025，Vol.38Issue(5)：826-832,7.DOI:10.3969/j.issn.1004-1699.2025.05.009
改进小波阈值降噪在涡轮叶片裂纹检测中的应用
Application of Improved Wavelet Threshold Denoising in Cracks Detection in Turbine Blade
摘要
Abstract
Due to curvature change of turbine blade surface,the lifting-off effect will produce large noise signals in eddy current(EC)detection.Meanwhile,the signal has local oscillation and distortion after denoising by traditional wavelet hard and soft threshold func-tions.Hence,a three-parameter adaptive wavelet threshold denoising algorithm is proposed to add a hierarchical correction factor to the threshold matrix,and each regulatory factor is optimized based on gray wolf algorithm(GWO).The denoising is simulated and compared by adding Gaussian noise to the Bumps test signal.After the application of improved threshold function,the signal-to-noise ratio(SNR)of denoising is increased by 76.94%,and the root mean square error is reduced by 75.90%.The signal of EC detection of turbine blade surface crack is analyzed,and the SNR of improved threshold denoising is up to 14.15 dB.In addition,support vector machine(SVM)method is used to extract the features of the detected signals after denoising,and the classifier accuracy reaches 93.3%,which is 195%higher than that before denoising.The experimental results show that the improved denoising algorithm is better than the traditional threshold denoising,effectively realizing the feature extraction and cluster analysis for EC detection signals in turbine blade crack.关键词
涡轮叶片/涡流检测/小波阈值降噪/灰狼优化算法/支持向量机Key words
turbine blades/eddy current testing/wavelet threshold noise reduction/GWO/SVM分类
计算机与自动化引用本文复制引用
王一乾,郭建城,胡明慧..改进小波阈值降噪在涡轮叶片裂纹检测中的应用[J].传感技术学报,2025,38(5):826-832,7.基金项目
国家重点研发计划项目(2018YFC1902404) （2018YFC1902404）
