Empirical investigation of acoustic emission signals for valve failure identification by using statistical method

This study proposed to employ acoustic emission (AE) signal to detect valve abnormalities in reciprocating compressor. The AE signals obtained are post-processed by using wavelet packet transform (WPT) to decompose the signals into different frequency ranges. To examine valve problems, root-mean-square (R.M.S) values of signals computed at each frequency and time segments are compared among three different valve conditions. The analysis of variance (ANOVA) test is conducted to investigate the effect of different valve conditions on the R.M.S value at its corresponding time and frequency segments. Finally, a post hoc test named Tukey comparison test is performed to identify time-frequency segments best representing the valve condition. It is believed that valve problems can be identified effectively by monitoring the R.M.S values at these segments.