热障涂层厚度的脉冲涡流检测反演方法

Inversion Method of Pulsed Eddy Current Detection for Thermal Barrier Coatings Thickness

针对热障涂层厚度的高精度测量,提出一种脉冲涡流检测反演方法。该方法提出两阶段处理技术对脉冲涡流检测信号进行预处理,采用主成分分析法提取预处理信号的主成分分量,最后构建BP神经网络实现涂层厚度预测。通过COMSOL建模仿真实验证明两阶段处理方法可有效地减小提离效应并区分厚度特征的变化,基于主成分分析的特征提取方法能够对陶瓷层和粘结层厚度变化进行分类识别。结果显示该方法对陶瓷层厚度计算的平均相对误差约为0.4%,对粘结层厚度计算的平均相对误差约为2.6%。可见,该方法对热障涂层厚度的反演精度较高。

To measure thermal barrier coatings thickness at the high precision,a pulsed eddy current detection inversion method is proposed.A two-stage processing technique is proposed to preprocess the pulsed eddy current detection signals,and the principal component of the preprocessing signal is extracted by principal component analysis method.Finally,BP neural network is constructed to predict the coatings thickness.The COMSOL modeling and simulation experiment proves that the two-stage processing method can effectively reduce the lift-off effect and distinguish the variations of thickness characteristics.The feature extraction method based on principal component analysis can classify and identify the thickness variations of ceramic layer and bonding layer.The results show that the average relative error of this method is about 0.4%for the thickness of ceramic layer and about 2.6%for the thickness of bonding layer.It can be seen that the inversion accuracy of thermal barrier coatings thickness by the mentioned method is higher.