Sparse Feature Extraction And Classification Identification Of Eddy Current Detection For Conductive Structure Defects

As a kind of non-destructive testing technology,eddy current testing technology provides a highly accurate and accurate defect detection method for conductive structural material,which improves the reliability of conductive structure material in practical application.The method of extracting and classifying the defective feature is the key to the application of the eddy current detection of the conductive structural material.The method of feature extraction of the defective signal is studied to quantitatively classify and evaluate the shape,size and material of the defect.It has always been the difficulty of the eddy current detection inverse problem.In this paper,conductive material as the object of eddy current testing,carried out for the conductive structure of eddy current detection of defective signal sparse feature extraction method and defect quantification classification recognition research,it has important research significance and application value.In this paper,based on the theory of electromagnetism and eddy current testing,the key techniques and mathematical models of eddy current testing are analyzed and studied.The finite element model of eddy current testing was established by ANSYS finite element simulation software.The finite element analysis of eddy current defect was carried out for different defect size and different detection probes.The finite element analysis was carried out.The results are related to the relationship between the defect type and the change of the probe form,and the variation rule of the test result brought by the change of each factor is summarized.A new method of Lagrangian multiplier K-SVD signal extraction based on sparse signal processing is proposed for the eddy current detection signal of conductive structural material.The eddy current testing experiment is carried out with aluminum alloy material as the measured object.The different defect length and the depth of the defect are set,and the detected impedance signal is obtained by the condition of changing the excitation frequency.The detection signal of different defect condition and different excitation frequency is extracted by using the Lagrangian multiplier K-SVD algorithm The results show that the Lagrangian multiplier K-SVD algorithm is feasible and efficient in the feature extraction of the defective signal,and the variation of the impedance of the probe generated by different defects and excitation frequency is calculated by calculating the characteristic distribution result by the above conditions.The SVM model was established,and the SVM model was optimized for the defects of the eddy current of the aluminunm alloy.The SVM classification and identification method was used to quantify and identify the defect characteristics by different defect size and different excitation frequencies.Analysis and Comparison of Principal Component Analysis and Lagrangian Multiplier K-SVD Algorithm in Feature Extraction and Classification Recognition,and verifying the Superiority of Lagrangian multiplier K-SVD Algorithm.