Identification of wave impacts and separation of responses using EMD

Publication Type:
Journal Article
Mechanical Systems and Signal Processing, 2021, 151
Issue Date:
Filename Description Size
j_ymssp_2020_107385.pdf9.32 MB
Adobe PDF
Full metadata record
Various marine and off-shore structures experience rare extreme, and frequent substantial wave impacts, and these impacts have significant implications respectively for the ultimate strength and fatigue life of these structures. Likewise, many other structures experience forms of impact among their environmental loads that are critical for their design. One way of exploring these impacts is to study the vibration signal of the structure. However, due to the complexity of the non-stationary signal, the wave impacts on marine and off-shore structures are difficult to analyse. Using signals acquired during trials of a high-speed catamaran as a case study, this paper proposes employing the Empirical Mode Decomposition (EMD) to detect, identify and characterise significant wave impacts (known as ‘slams’), which is shown to be significantly better at classifying events than traditional methods of slam detection. With the application of EMD, the vibration signal is decomposed into many components that may be grouped into three categories: (1) the rigid-body response comprising two parts: the quasi-static response to the underlying wave spectrum (hydrostatic) and rigid-body resonance (hydrodynamic); (2) the elastic structural response (hydroelastic); and (3) local high frequency vibrations and/or noise. It is the second category that responds significantly to wave impact loads. By identifying the gap between the rigid body and structural resonant frequencies, a threshold is established to automate the physically rational separation of this hydroelastic response. Wave impact detection methods are then applied to this separated component, showing 94% true positive classification, compared with 64% for a recently published slam detection method that uses traditional filtering applied to the whole signal. Further, it allows for more targeted subsequent characterisation of the impact response. Thus, this paper concludes that EMD is an effective method to detect and identify wave impacts.
Please use this identifier to cite or link to this item: