Field implementation of low-cost RFID-based crack monitoring using machine learning

As civil infrastructure continues to age, the extension of service life has become a financially attractive solution due to cost savings on reconstruction projects. Efforts to increase the service life of structures include nondestructive evaluation (NDE) and structural health monitoring (SHM) techniques. Nonetheless, visual inspection is more frequently used due to high equipment cost from other techniques and federal biennial inspection requirement. Recently, low-cost Radio Frequency Identification Devices (RFID) have drawn attention for crack monitoring; however, it was yet to be implemented in the field. This paper presents a crack monitoring algorithm using a developed RFID-based sensing system employing machine learning under temperature variations for field implementation. Two reinforced concrete buildings were used as testbeds: a parking garage, and a residential building with crumbling foundation phenomenon. An Artificial Neural Network (ANN)-based crack monitoring architecture is developed as the machine learning algorithm and the results are compared to a baseline model. The results show promise for field implementation of crack monitoring on building structures.