Defects identification using the improved ultrasonic measurement model and support vector machines

With a combination of the improved ultrasonic measurement model (IUMM) and support vector machines (SVM), a novel method to identify inclusions and cavities in metallic materials using scanning acoustic microscopy is proposed. In the IUMM, a hybrid model of Born approximation and Kirchhoff approximation is developed to calculate the far-field scattering amplitude of cavities, which improves the accuracy in phase and amplitude of the predicted pulse-echo signals of defects. The SVM classifier, with the amplitude and peak frequency of the predicted echo signals as major features, is applied to distinguish inclusions and cavities. The experimental result shows that the echo signals predicted by the proposed IUMM are more accurate than conventional UMM in amplitude and frequency. The SVM classifier, with the predicted signals as the training set, enables the identification of inclusions and cavities in metallic materials successfully. This work improves the performance of SAM in the identification of internal defects in metallic materials and realizes the intelligent analysis of ultrasonic signals.