Model-based Intelligent Recognition for Aluminum Plate Seam Defects

Abstract In order to study the application of nonlinear ultrasonic in the quantitative identification of defective aluminum plate, different depth cracks are machined on the aluminum alloy plate with a thickness of 10 mm by wire cutting to simulate the defects in the plate. The normal and defective aluminum plates are selected to establish the experimental model, and the continuous wavelet transform (CWT) is used to extract the characteristic parameters of the aluminum plate nonlinear ultrasonic signal. The dimensions of the data are reduced by principal component analysis (PCA), and the principal component with the top three contribution rate are selected as the characteristic value. Finally, the support vector machine (SVM) algorithm is used to analyze the aluminum alloy plate state and classify the defect signal. The experimental results show that the feasibility of nonlinear ultrasonic signal recognition of aluminum plate defects is verified by combining principal component analysis and support vector machine model.