Digital twins for compliant and elastic components

Abstract

Novel research fields are gradually finding their place within the Industry 4.0 ecosystem. This is noticeable with compliant mechanisms, soft robots, and other instances of compliant and elastic components. The complex interaction they exhibit between kinematics and elastic deformation presents a significant obstacle for their cyber-physical integration: the development of physically accurate models with real-time capabilities. This has resulted in their slower integration into the 4.0 ecosystem compared to other more mature technologies. This work addresses the design of a cyber-physical integration framework for compliant components from a modeling perspective. This framework can be utilized in the design of digital twins and other mixed-domain applications. To accomplish this, reduced order modeling is employed for the creation of efficient virtual representations, and computer vision is used for integration with the physical domain via sparse structural features tracking. The proposed cyber-physical integration framework is applied in the implementation of the digital shadow of a compliant joint and in an exploratory study of fusing experiments and simulations of mechanical metamaterials. This work contributes to the development of a data-driven methodology for recovering the parameter-dependence of a reduced basis approximation, initially having a single parameter and later extended to multiple parameters. This is validated numerically as well as proven theoretically. Additionally, a methodology is defined for acquiring sparse structural information of mechanical metamaterials using a vision-based approach for the visual tracking of structural features. Furthermore, this work studies the imposition of essential boundary conditions on nonlinear reduced basis models. Although this part is still at an exploratory stage, it might be a promising alternative for modeling compliant and elastic components subjected to large deflections, a topic still under development in the literature in which a generalizable solution has not yet been proposed. Control and monitoring applications for compliant mechanisms could be the first to benefit from this modeling strategy. However, the nonlinear reduced basis models yielded execution times far from real-time, which would require leveraging other technologies like hardware acceleration. Future work is encouraged to explore the utilization of static condensation and multipoint constraints for creating a modular framework for modeling larger-scale systems of compliant components. Also, to conduct further experimental validation and to develop novel cyber-physically integrated applications based on the proposed framework.