Study of Feature Extraction Method in Ultrasonic Phased Array Testing for Long-Distance Pipeline

Abstract

Though ultrasonic phased array technology is more efficient than traditional manual ultrasonic testing method, automatic flaw classification is a challenge and still hasn’t been well solved. Whether the representative features can be extracted from each type of ultrasonic flaw signal is a key to influencing the accuracy rate of automatic flaw classification. In this paper, second generation wavelet transform (SGWT) is proposed as a flaw feature extraction method, having the advantages of high computation speed, simple structure and occupying less memory. After introducing the principle of SGWT, the SGWT-based feature extraction algorithm is analyzed. Separability measure based on Euclidean distance is introduced as the evaluation criterion to assess flaw feature extraction performance. For comparison, first generation wavelet packet transform (WPT), a common feature extraction method, is also adopted to extract flaw feature. The experiment result is indicated that the classification performance of SGWT-based feature extraction algorithm is improved than WPT-based feature extraction algorithm, and the classification speed of the former is almost two times of the latter, which is valuable for automatic flaw detection and classification of pipeline girth weld.