Structural health monitoring has increasingly become a hot research topic for engineering structures. While much improvement has been made in advances of sensing, communication and computer technologies, identification of structural parameters based on vibration signals is a promising but difficult task due to the ill-conditioned nature of inverse analysis. For offshore structures, the challenge is even greater as they are usually very large and subjected to severe environmental loads including dynamic wave load. This keynote focuses on offshore jack-up, a type of mobile self-elevating drilling unit used in the oil and gas industry. A jack-up rig typically comprises a buoyant hull supported by three lattice legs, each resting on a large inverted conical footing called spudcan. As jack-up peration moves into deeper waters and harsher environments, there has been an increasing demand to model the jack-up structure more accurately by accounting for spudcan fixity (i.e. soil stiffness at spudcan footing). Due to the large size of jack-up rig, a substructure approach is essential so as to reduce the effort of forward analysis and inverse analysis. In view of the difficulty in measuring wave forces, a predictor-corrector procedure is incorporated to compute the input force instead as identification progresses based on an improved genetic algorithm method recently developed by the authors. The performance of the proposed strategy in identification of sudcan fixity is demonstrated by numerical study, using a jack-up operating in the North Sea as an example.