基于1DCNN的碳纤维复材裂纹损伤涡流检测方法OA

To address the difficulty of locating and quantifying subsurface damage in eddy current testing of carbon fiber composite materials,the research group proposed a method based on a 1-Dimensional Convolutional Neural Network(1DCNN)for eddy current testing.By constructing a benchmark dataset covering 9-dimensional variables,including scanning position and angle,and employing a multi-modal self-selection fusion technique to integrate 4 signal modalities,an end-to-end 1DCNN model from eddy current signals to defect parameters was established.Quantitative evaluation results show that this method achieves an Mean Absolute Error(MAE)of 0.36 mm for subsurface crack damage localization and 0.02 mm for depth quantification evaluation.This study provides a practical intelligent solution for the nondestructive testing of composite material structures.