Application of Acoustic Emission and Support Vector Machine to Detect the Leakage of Pipeline Valve

Abstract

In order to effectively detect the leakage of the pipeline valve in operation sate, a method was proposed based on acoustic emission (AE) theory and support vector machine (SVM) model, firstly, the acoustic emission testing platform was setup, and then, AE testing for valve internal leakage under test platform was performed, and the root mean square (RMS), average signal level (ASL) of the time domination and peak value of the frequency domination were as eigenvectors for the SVM model. Finally, the SVM model for the detection of leakage of pipe valve was established through the training and testing eigenvectors, and the abilities of the kernel functions were evaluated. Results show that the method based on RBF kernel function is workable and effective for the leak detection of pipe valve with the sensitivity of 92.5%, the specificity of 100%, and the accuracy of 96.25%.