Damage localization for offshore platform by neural networks

In this paper, a damage localization approach for offshore platform by artificial neural networks is proposed. The members of offshore platform structure are classified and separated into several layers. The decision system for the kind and layer of damaged members is established using the back-propagation networks. When inputting the change rate of normalized frequency into the decision system, the kind and layer of damaged members are determined. The decision system for the location of damaged members is established using the probabilistic neural networks. When inputting the normalized damage-signal index into the decision system, the location of damaged member is determined. Numerical simulations demonstrate that the approach can localize the damage of offshore platform with good robustness.