Output-only modal identification of full-field time-domain data for heated hybrid hollow sandwich structures

In order to enhance the efficacy of decomposing output-only vibration response signals and conducting adaptive signal analysis, this paper introduces a novel non-parametric modal identification methodology tailored for multi-sensor signals. This approach synergistically integrates successive multivariate variational mode decomposition (SMVMD) and high-speed three-dimensional digital image correlation (3D-DIC) technology. A thermal modal testing system was meticulously established, operating under a nonuniform temperature condition of 900 °C, enabling the acquisition of thermal modal parameters for the hybrid hollow sandwich structure (HHSS) both at room temperature and across diverse nonuniform temperature fields. Utilizing 3D-DIC, the method systematically extracts various positions of the test structure along with the corresponding degrees of freedom (DOFs), thus facilitating the creation of a comprehensive dataset. The Q-learning algorithm is employed to simulate the divisional local radiant thermal environment resulting from the combination of a finite element model (FEM) involving a quartz lamp array and the HHSS. Conclusively, a multistate nonuniform temperature field model updating methodology is employed to refine the HHSS, yielding modal frequency errors within a 7 % margin in the updated dynamic model and concurrently enhancing the accuracy of the FEM. The outcomes reveal that exposure to elevated temperatures induces permanent thermal buckling deformation in the HHSS under free-free boundary conditions, particularly evident in regions devoid of a honeycomb core. Notably, the deformation directions in the heated and nonheated regions exhibit inconsistency. These findings provide a reference basis for the dynamic performance analysis and adaptive modal identification of lightweight structures under a non-uniform temperature field.