Integrity monitoring of offshore platforms using artificial neural nets

Abstract

This paper explores the possibility of using artificial neural Nets (ANNs) for monitoring the integrity of offshore structures. The methodology consists of comparing measured rms displacement responses at different elevations of an offshore structure, for a given sea state, with a datum set, obtained using a number of trained feedforward ANNs. Each ANN, given the significant wave height and average wave time period that characterise a sea state, provides structure response estimates for a pre-defined damage state. These estimated responses are suitably sealed to generate the datum set. If the measured response pattern does not match the undamaged state ANN response pattern, the structure is damaged. The region of damage is given by the damage state whose ANN response pattern matches the measured response pattern. In this paper analytically determined rms displacements from a simulation study are used, instead of measured displacements from the platform structure, to train the ANNs.