Decentralized digital twins for structural health monitoring

• Applicant: Professor Dr.-Ing. Kay Smarsly
• Subject Area: Structural Engineering, Building Informatics and Construction Operation
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 542312939

Amid the digital transformation, structural health monitoring (SHM) is increasingly relying on wireless sensor networks that transmit structural response data to central servers for offline data analysis, which, however, is associated with several problems, as (i) the requirements for data storage and management may be difficult to fulfil, (ii) analyzing large amounts of data may require powerful database systems and complex algorithms, and (iii) the wireless data transmission may cause issues of data security and data loss. Therefore, research has focused on decentralizing data analysis, via embedding data-driven models into the microcontrollers of wireless sensor nodes, to enable autonomous data analyses on the sensor nodes. However, embedded data-driven models yield limited information on the structural condition, as compared to the information extracted by centralized offline data analyses that combine structural response data with physics-based models, such as finite element models. As a result, a decentralized data analysis approach that (a) overcomes the problems of centralized data analysis but (b) yields rich information on the structural condition is necessary. This project aims to develop a methodology for decentralized data analysis in wireless SHM systems using information-rich numerical models, where digital twins (communicating with each other) are distributedly embedded into the microcontrollers of the wireless sensor nodes. The digital twins will be first segmented into “partial digital twins”, and each partial digital twin will be embedded into one wireless sensor node and used for local data analysis. In this project, digital twins will be represented by finite element models that approximate the structural condition. As a result, the partial digital twins essentially will be formulated as “partial finite element models”, i.e. finite element models segmented into substructures with internal degrees of freedom and interface degrees of freedom. The partial digital twins will be tailored to the characteristics and computational capabilities of wireless sensor nodes, in terms of geometry, mesh refinement, and computational cost. The expected outcome of the project is a methodology for decentralizing digital twins for SHM that advances the level of information on the structural condition currently obtained by state-of-the-art embedded data-driven models, enabling wireless sensor nodes to describe the structural condition in detail. The project is expected to enhance the efficiency and robustness of SHM, thus fulfilling the needs of the increasingly digitalized landscape of structural maintenance. The outcome of the project will align SHM with the latest concepts in digital transformation, building upon the emerging paradigms within Industry 4.0.

• DFG Programme: Research Grants