Chillón Geck, CarlosCarlosChillón GeckAl-Zuriqat, ThamerThamerAl-ZuriqatElmoursi, MoayedMoayedElmoursiDragos, KosmasKosmasDragosSmarsly, KayKaySmarsly2024-09-122024-09-122024-07-0111th European Workshop on Structural Health Monitoring (EWSHM 2024)https://hdl.handle.net/11420/49048Artificial intelligence (AI) algorithms have proven effective in implementing sensor fault diagnosis (FD) for wireless structural health monitoring (SHM). However, FD models based on AI are computationally expensive and require large amounts of raw sensor data to be transmitted to centralized servers. This paper proposes a decentralized framework for sensor fault diagnosis in wireless SHM systems based on the concept of Artificial Intelligence of Things (AIoT). Within the decentralized framework, FD models are embedded into wireless sensor nodes to ensure that the data collected from engineering structures is fault-free. Thus, only the condition of SHM systems, instead of raw data, is transmitted from SHM systems to centralized servers via Internet-of-Things communication. To validate the decentralized framework proposed in this paper, an SHM system is implemented using (i) a portable main station containing the FD models and (ii) four tailor-made wireless sensor nodes equipped with microcontrollers and accelerometers deployed on a test structure. The results of the validation tests show that the SHM system successfully collects acceleration data and diagnoses, in real-time, sensor faults that are inserted into the sensor nodes. In future work, the decentralized framework and the SHM system presented in this paper may be deployed on a bridge for structural condition assessment, while ensuring early detection of sensor faults.en1435-49342024NDT.nethttps://creativecommons.org/licenses/by/4.0/Artificial Intelligence of Things (AIoT)Internet of Things (IoT)sensor fault diagnosisStructural health monitoring (SHM)Technology::624: Civil Engineering, Environmental Engineering::624.1: Structural EngineeringConference Paper10.15480/882.1328810.58286/2957710.15480/882.13288Conference Paper