Metamodels based on ARMAX time series prediction and application on structural digital twins

Authors:

Abstract
Prediction of structural response, stress concentration and eventually fatigue is important for safety and life extension of floating platforms and wind turbines. A method based on reduced order models that are based on parametric finite element analysis, artificial intelligence and time series prediction is proposed here. In particular, the ARMAX family of time series prediction method has been trained and tested for the prediction of the quantities of interest, based on simple and complex finite element models of structures subjected to wave and earthquake loadings. After calibration, prediction of the quantities of interest at non accessible points is possible, in the sense of virtual sensors. The resulting system is able to assist structural health monitoring and support predictive maintenance tasks.

This publication has 12 references indexed in Scilit: