Integrated ARMA model method for damage detection of subsea pipeline system

Chunxiao Bao, Hong Hao, Z. Li

    Research output: Contribution to journalArticle

    44 Citations (Scopus)


    An integrated ARMA model algorithm is developed in this study for the structural health monitoring (SHM) of subsea pipeline system. In which, the partition and normalization procedure is firstly employed in signal pre-processing to remove the influence of various loading conditions, the auto-correlation function of the normalized signal is utilized as a substitute of analysis input to overcome noise effect and avoid the bias in Autoregressive Moving Average (ARMA) model fitting caused by noise disturbance as well. Then, Partial Auto-correlation Function (PAF) method is employed in building optimal ARMA model. With which Autoregressive (AR) parameters serving as damage feature vector, a damage indicator (DI) based on the Mahalanobis distance between ARMA models is defined for damage detection and localization. Dynamic vibration of subsea pipeline system under ambient excitations is numerically simulated in ANSYS software and the acceleration responses of pipeline in various damage cases are analyzed utilizing the proposed integrated method. In numerical study, the finite element (FE) model of a subsea soil-pipeline-fluid system is developed and the undersea hydrodynamic force acting on the pipeline is derived based on Spectral Analysis Method. Finally, the proposed method is proved to be robust and very sensitive to damage. It provides accurate identification of damage existence and damage locations with high time efficiency. Therefore, it can be used for the online SHM of subsea pipeline structures and other civil structures. © 2012 Elsevier Ltd.
    Original languageEnglish
    Pages (from-to)176-192
    JournalEngineering Structures
    Publication statusPublished - 2013

    Fingerprint Dive into the research topics of 'Integrated ARMA model method for damage detection of subsea pipeline system'. Together they form a unique fingerprint.

    Cite this