Determination of Characteristic Frequency Segments of Acoustic Emission Signal for Valve Leakage Detection in Reciprocating Compressor

Abstract

Acoustic emission (AE) technique is often employed by condition monitoring personnel to detect abnormalities in moving components of machines. Due to its low signal-to-noise ratio at high-frequency range, it often requires certain signal processing technique to extract valuable information from machine parts. This study intends to obtain the characteristic frequency segments of AE signals that correspond to the physical movement of valves in reciprocating compressor. It involves acquisition of AE signals at various simulated valve conditions and rotating speeds, decomposition of these signals through wavelet packet transform (WPT), and computation of crest factor (CF) of WPT coefficients at a specific crank angle. The characteristic frequency segments that indicate valve problems are often accompanied by an extra high or low CF value. By relating the CF value to valve events, the condition of valve can be predicted. It is hoped that this study can provide a methodology to obtain valve information effectively and efficiently while reducing unwanted and overloading information of AE signals.