Digital twin framework for structural health monitoring using PWT-excited Lamb waves

Non-destructive structural health monitoring (SHM) employs techniques that allow for the continuous evaluation of structural integrity without causing harm to the structure itself. By utilising methods such as Lamb wave propagation and detection using piezoelectric wafer transducer (PWT), these systems enable the early detection of defects or damage, ensuring the safety and longevity of critical infrastructure. PWTs are employed to generate and detect guided waves, such as Lamb waves (LW), within the structures. These waves propagate through the structure, and the responses can be analysed to identify the presence of damage. Nonetheless, damage identification via damage-sensitive features derived from diverse signal processing techniques may prove ineffective in real-world settings characterised by several uncertainties. To address these challenges, a digital twin model that accommodates these uncertainties in SHM using PWAS is presented. The bonding between the sensor and the structure can significantly impact SHM applications where partial contact loss may lead to incorrect conclusions, potentially being misinterpreted as damage within the structure. A data-driven approach is implemented to develop a digital twin (DT) model that distinguishes between healthy and damaged states. Experimental studies are conducted with a wafer-type sensor coupled to an elastic plate, and a laser vibrometer is used as a receiver to analyse the AO mode behaviour at different points within a 30 cm radius surrounding the sensor. Mode tuning curves for a frequency range from 13 kHz to 260 kHz are obtained, with the amplitude found to vary with bonding conditions. The obtained response is then used to train the digital twin model, which can be further tested with damage scenarios.