Nonstationary analysis of vibration signals of diesel engine based on LCS

The detection of nonstationarity and nonlinear dynamics of nonstationary chaotic time-series is a challenge issue, since the conventional methods are all based on the assumption that the objective time-series is stationary. In this paper, we propose a new method for analyzing the nonstationary chaotic time-series, i.e. the phase space reconstruction is directly done on the nonstationary time-series through Hilbert transform, which retains both the nonstationarity and the nonlinear dynamics of the original time-series. Then, the nonstationarity and the nonlinear dynamics are measured via Lagrangian coherent structures (LCS), and a new algorithm for calculating the maximum finite-time Lyapunov exponent (FTLE) is presented. Simulation tests are conducted on a set of vibration signals of diesel engine. Results indicate that the proposed method can provide new features to efficiently identify the nonstationarity and the nonlinear dynamics of nonstationary chaotic time-series.