Turbine Rotor Vibration Faults Diagnosis based on Wavelet Packet Analysis and the Largest Lyapunov Exponent

According to the four typical fault signals of turbine rotor vibration, including rubbing, loosening, misalignment and mass unbalance which are collected from the Bently experiment set, the method which combines wavelet packet and the largest Lyapunov exponent is adopted to diagnose the faults. First, use wavelet packet analysis for filtering the fault signals, extracting useful signal frequency segments from original signals. Then calculate the largest Lyapunov exponent of the signals which is adopted as the basis of diagnosis. The results indicate that wavelet packet analysis is of good ability of filtering and extracting of nonstationary signal; the largest Lyapunov exponent of vibration time series can distinguish different fault conditions, and is of good discriminability in turbine rotor faults diagnosis.