LAUSR

Abstract Galvanized steel pipes are widely used for indoor gas distribution. Leakage in the pipeline system is prone to occur in screw thread connection as opposed to tube itself. Early detection of small leak is of great significance to ensure the safety and comfort in doors. This work presents an experimental investigation on AE based small leak detection of galvanized steel pipe due to screw thread loosening. The waveform, frequency and energy signatures of the AE signals are first extracted and compared. Regarding the fact that the small leak signals lack obvious characteristics and are easily submerged in background noise, a pattern recognition method based on support vector machine (SVM) is employed for leak detection. Through training and testing on the experimental data, it is verified that the algorithm based on SVM with RBF kernel function is of the highest accuracy and efficiency with a 1.9% false alarm rate at most.