Virtual Sensing and Sensor Fusion Techniques for Strain Estimation Based on Acceleration Data and a Multiple-Input FIR Filter

Structural Health Monitoring (SHM) is employed to evaluate the condition of a structure based on the analysis of measurement data. A significant challenge in SHM is the inability to perform direct measurements at every critical point within a structure. To overcome this limitation, virtual sensor approaches are developed to estimate the measurement values at positions not equipped with real sensors. In the case of lifetime assessment of dynamically loaded structures, such as wind turbine support structures or bridges, it is beneficial to estimate strain at different positions based on acceleration data. Enhancing prediction accuracy necessitates the fusion of data from multiple sensor sources, enabling a reliable assessment of structural integrity. This contribution presents an innovative approach utilizing Multiple-Input Finite Impulse Response (FIR) filters, which are identified using a data-driven approach with both Finite Element (FE) model data and empirical measurements. The method is validated and compared to the modal decomposition and expansion (MDE) approach on data from an experiment performed using a laboratory cantilever beam, demonstrating its potential for widespread application in SHM. Furthermore, the results highlight the benefits of integrating virtual sensing and sensor fusion for improved monitoring and maintenance of critical infrastructure.