Digital twin-based structural health monitoring by combining measurement and computational data: An aircraft wing example

Abstract

Digital twin is a concept that utilizes digital technologies to mirror the real-time states of physical assets and extract the hidden yet valuable information of physical assets for optimization, decision-making or scheduling. By combining measurement and computational data, this paper presents a digital twin-based structural health monitoring framework of physical assets. The process for building the measurement-computation combined digital twin (MCC-DT) involves four steps. First, an artificial intelligence-driven load identification method combining measurement and computational data is employed to recognize the loads applied on physical assets. Two approaches were proposed to realize load identification, based on single fidelity surrogate models and deep learning techniques, respectively. Second, multi-fidelity surrogate (MFS) models are applied to improve the accuracy in the MCC-DT. Two routes for implementing the MFS models are introduced and the advantages and shortcomings of both are analyzed. Third, an online rainflow counting algorithm is developed to calculate the degradation of the physical assets. The main advantage of the algorithm is that it can provide a near real-time estimation for the damage accumulated of physical assets. Finally, the data generated from the first three steps can be fused into a three-dimensional scene using Web graphics library to provide an intuitive view of the MCC-DT. To describe the implementation details of the framework and verify its applicability and effectiveness, the MCC-DT was established using an aircraft model as an example.