Pipeline Defect Classification by Using Non-Destructive Testing and Improved Support Vector Machine Classification

CATEGORY: NON DESTRUCTIVE TESTING
International Journal of Engineering and Innovative Technology (IJEIT), Volume 2, Issue 7, January 2013
Pipeline Defect Classification by Using Non-Destructive Testing and Improved Support Vector Machine Classification
Muhsin Hassan, Rajprasad Rajkumar, Dino Isa, Roselina Arelhi
Department of Electrical and Electronic Engineeering, Faculty of Engineering University of Nottingham, Malaysia Campus, Jalan Broga 43500, Semenyih Selangor
Abstract
This paper deals with the efforts to improve classification accuracy of non-destructive testing (NDT) techniques by implementing hybrid of Kalman Filter (KF) and Support Vector Machine (SVM) in classifying corrosion depth.
The main emphasis is on the improvement of accuracy for SVM classification in noisy environment with the help of Kalman Filter. Long range ultrasonic testing will be employed, where a ring of piezoelectric transducers are used to generate torsional guided waves. Various defects such as cracks as well as corrosion under insulation (CUI) will be simulated on test pipe. The machine learning algorithm known as the SVM will be used to classify transducer. The classification performance of SVM was exceptional, showing a facility to detect defects at different depths as well as for distinguishing closed spaced defects however this does not perform well with noisy data. This paper proposes the idea of using Kalman filter to filter out the noise and improves the classification accuracy of SVM classification on noisy data. A Discrete Wavelet Transform (DWT) with SVM application will be used as benchmark for comparison purposes.
Free Full Text Source: http://www.ijeit.com/vol%202/Issue%207/IJEIT1412201301_13.pdf