Diagnosis of Pipeline Condition Based on ELMD and AOK Spectral Entropy

This paper proposes a pipeline condition diagnosis method based on ELMD and AOK spectrum entropy. The ensemble local mean decomposition (ELMD) algorithm can solve the problem of modal aliasing in the local mean decomposition (LMD) algorithm, and has little endpoint effects and few iterations. Adaptive optimal kernel (AOK) time-frequency distribution has the advantages of high time-frequency resolution, good time-frequency clustering, and the ability to suppress cross-interference. The AOK spectral entropy parameters are extracted from the PF component of the ELMD decomposition in order to describe quantitatively the time-frequency characteristics of the signal. Simulation results show that the three operating conditions including the normal operation of the pipeline, the pipeline leakage and the pipe knocking can be accurately identified by ELMD and AOK spectral entropy parameters and the overall recognition accuracy is 93.3%.