Trend analysis approach based on empirical mode decomposition

An empirical mode decomposition (EMD) based approach for trend analysis of mechanical running condition is investigated. The results show that the approach can extract the trend of a signal more precisely, comparing with the traditional trend analysis methods, such as least square approximation and low-pass filtering. The presented approach has been applied to turbine fault diagnosis in an oil refinery, and the results indicate that it can obtain the trend of mechanical running condition effectively, then evaluate machine health condition and diagnose machine fault accurately.