Digital twin-enabled structural real-time monitoring for blended-wing-body underwater gliders

Abstract

As innovative marine systems, Blended-wing-body underwater gliders (BWBUGs) have become crucial components in maritime operations. However, during autonomous missions beyond shore-based support infrastructure, BWBUGs face significant challenges from complex marine environmental conditions. Thus, real-time structural monitoring and prediction become essential for ensuring operational stability. To address these requirements, this study develops a digital twin-enabled framework for BWBUG structural monitoring, integrating three key components: data generation and collection, sensor layout optimization, rapid prediction and visualization. Specifically, the framework implementation initiates with baseline dataset generation employing numerical simulations and experiments, with numerical models calibrated iteratively against experimental measurements. Subsequently, the Kriging-assisted discrete global optimization (KDGO) algorithm is employed for optimal strain sensor placement. The final phase implements a predictive model between sparse sensor inputs and full-field strain data, combining isometric mapping (Isomap) dimensionality reduction with radial basis function (RBF) surrogate modeling. Concurrent virtual visualization techniques are implemented for analysis of structural responses. Experimental validation confirms the framework's effectiveness in BWBUG structural real-time monitoring, and its data-driven architecture enables adaptation to other marine equipment requiring real-time monitoring.