Neural Network Application for Wide Band Fatigue Damage of Floating Offshore Wind Turbine Mooring Line

초록

It is widely known that there have been challenging tasks to accurately predict wide band spectral fatigue damage in marine structures. Though mooring dynamic analysis in time domain usually requires considerable amount of cost, the most accurate wide band fatigue damage can be obtained from time domain analyses. Among many candidates to reduce the costs, artificial neural network is known to be very effective one. Artificial neural network (ANN) model can provide fatigue results for a given environmental condition without performing any numerical calculation as long as the ANN is sufficiently trained. In this study, OC4 floating offshore wind turbine is considered as a target structure where three catenary mooring lines are assumed, and most environmental loading data are collected from the Jeju offshore area. ANSYS/AQWA (2015) is used to carry out hydrodynamic simulations for the platform with which the ANN is trained. In order to validate performance of the trained ANN, hydrodynamic simulations for new environmental conditions are carried out and the simulation results are compared with the results from the ANN. It is proven that two results are in an excellent agreement in terms of tension range distributions of a mooring line.