ANN-based surrogate models for the analysis of mooring lines and risers

• This work presents new surrogate models based on artificial neural networks (ANNs).
• The goal is to obtain striking reductions in CPU time compared to FE analyses.
• ANN associated to Nonlinear AutoRegressive model with eXogenous inputs (NARX).
• Case studies determine best model configurations and evaluate their performance.

This work presents a new surrogate model based on artificial neural networks (ANNs), comprising a rapid computational tool for the analysis and design of mooring lines and risers. The goal is to obtain results nearly as good as those provided by expensive finite element (FE)-based nonlinear dynamic analyses, with dramatic reductions in processing time. The procedure proposed here associates an ANN with a Nonlinear AutoRegressive model with eXogenous inputs (NARX). Differently from previous models based purely on exogenous inputs (i.e. the platform motions), the NARX model relates the present value of the desired time series not only to the present and past values of the exogenous series, but also to the past values of the desired series itself. Case studies are presented to determine the best configurations for the model, and to evaluate its performance in terms of accuracy and computational time.