Detection of damage in offshore jacket structure using artificial neural network

Abstract

With industrial growth over the years, geologists have begun to search for offshore sources of oil and natural gas. Most offshore oil-production platforms are jacket type. Offshore jacket structures consist of welded steel space frame. Tubular pipes are opted as they reduce hydrodynamic loads making them highly durable structures. However due to natural phenomenon like cyclones, earthquakes, etc. these structures can get adversely damaged. Apart from that the structure also degrades due to corrosion and fatigue due to the severe conditions it is always exposed to. In many cases physical inspection of the structure for checking damages is not possible due to unfavourable onsite conditions or lack of trained professionals. Hence there is a need to develop alternatives solutions. In recent years much research has been done in the application of Artificial Neural Networks (ANN) in Civil Engineering. ANN are computation systems that are inspired by biological neural systems. This paper aims at creating an ANN to identify damage in an offshore jacket structure using the modal parameters. The training set for the ANN is obtained by using a finite element software. The ANN is then tested using a test set and then it will be used to predict the structural damage in an offshore jacket structure.