COMBINING DISPLACEMENT AND ACCELERATION DATA FOR STRUCTURAL HEALTH ASSESSMENT

Traditional vibration monitoring systems often use accelerometers to record structural responses to various excitation sources, with changes in modal parameters indicating potential damage. However, displacement measurements offer more direct and valuable information about structural health, revealing operational conditions and detecting permanent deformations. Direct displacement measurement, though, is challenging and has several limitations tied to the technology used. For example, GNSS (Global Navigation Satellite Systems) receivers are less suitable for dynamic applications due to low sampling rates, while LVDTs (Linear Variable Differential Transducers) require a fixed base, which is often unavailable. Additionally, double integration of acceleration fails to provide accurate quasi-static or residual displacement measurements. An emerging approach to address the limitations of traditional monitoring systems is data fusion, which combines measurements from various sensors to improve accuracy. Typically, accelerations are integrated with data from strain gauges, GNSS, cameras, inclinometers, and radar. This paper explores the potential of data fusion for structural health monitoring by combining displacement data estimated through vision-based techniques with traditional accelerometer-based measurements. This combination leverages the accelerometer ability to measure very small vibrations with high time resolution and the vision-based approach capability to detect residual or quasi-static displacements. The paper presents results from merging vision-based displacement data with co-located accelerations measured on a laboratory scale frame. Data fusion, involving sensors with different sampling rates, is performed using a multi-rate Kalman filter.