Authors :
Lizbeth Kariza Gomes; Madhuraj Naik
Volume/Issue :
Volume 8 - 2023, Issue 3 - March
Google Scholar :
https://bit.ly/3TmGbDi
Scribd :
https://bit.ly/3Kkqjgn
DOI :
https://doi.org/10.5281/zenodo.7807345
Abstract :
Over the past few years with
industrialization has necessitated humans to consider
offshore resources of natural gas and oil. Fixed offshore
jacket structures constitute an important part of
offshore rigs. These structures are prone to damage due
to their long term exposure to the saline environment
which causes corrosion. Hence the health of these
structures have to be overseen periodically to
comprehend the extent of damage which in turn helps to
decide a suitable course of action for the maintenance of
the structure. Recently a lot of study had been carried
out on the use of Artificial Neural Networks (ANN) in
monitoring the health of a structure. ANN are a branch
of machine learning and works by imitating the human
brain. For this study, an offshore jacket was modelled in
a finite element based software in order to create the
training set. Damage was represented in the structure
by two different methods separately. First method was
by representing the damage through reduction of the
area of cross section of the main tubes and second
method was by using the reduction in elastic modulus to
represent the damage. Three different neural networks
were prepared for each method of damage
representation with different input parameter cases
namely modal frequency, modal frequency and eight
nodal displacements, modal frequency and twelve nodal
displacements. The optimum number of neurons in the
hidden layer was obtained for the respective case. Each
network was tested using a test set and output of the
networks were compared with the true value of the
damage. The results of the two methods of damage
representation were compared.
Keywords :
Artificial Neural Networks; Damage; Elastic Modulus; Modal Frequency; Offshore Jacket; Structural Health Monitoring
Over the past few years with
industrialization has necessitated humans to consider
offshore resources of natural gas and oil. Fixed offshore
jacket structures constitute an important part of
offshore rigs. These structures are prone to damage due
to their long term exposure to the saline environment
which causes corrosion. Hence the health of these
structures have to be overseen periodically to
comprehend the extent of damage which in turn helps to
decide a suitable course of action for the maintenance of
the structure. Recently a lot of study had been carried
out on the use of Artificial Neural Networks (ANN) in
monitoring the health of a structure. ANN are a branch
of machine learning and works by imitating the human
brain. For this study, an offshore jacket was modelled in
a finite element based software in order to create the
training set. Damage was represented in the structure
by two different methods separately. First method was
by representing the damage through reduction of the
area of cross section of the main tubes and second
method was by using the reduction in elastic modulus to
represent the damage. Three different neural networks
were prepared for each method of damage
representation with different input parameter cases
namely modal frequency, modal frequency and eight
nodal displacements, modal frequency and twelve nodal
displacements. The optimum number of neurons in the
hidden layer was obtained for the respective case. Each
network was tested using a test set and output of the
networks were compared with the true value of the
damage. The results of the two methods of damage
representation were compared.
Keywords :
Artificial Neural Networks; Damage; Elastic Modulus; Modal Frequency; Offshore Jacket; Structural Health Monitoring