ON THE USE OF HETEROGENEOUS , WIRELESS SENSOR NETWORKS FOR DAMAGE ASSESSMENT IN BRIDGES UNDER UNKNOWN EXCITATIONS

A multi-scale wireless sensor network was previously introduced by the authors and their collaborators, integrating data from a heterogeneous sensor array for a more robust and effective approach to damage detection. This multi-scale network concept was developed to improve power efficiency, minimize packet loss and latency, and eliminate synchronization issues through the use of decentralized analysis schemes and the activation of sub-networks only in the vicinity of suspected damage. Specific focus is given here to the vibration-based damage detection algorithms operating in the lower tier or micro-network, which do not require knowledge of the system input. Two approaches for decentralized system identification are evaluated: Method 1 is a transfer function-based approach, while Method 2 derives its damage sensitive feature directly from regressive model coefficients. Both use heterogeneous sensor measurements in lieu of measured inputs to the system. The utility of data fusion, spatially and temporally, within the upper tier or meso-network is also demonstrated as a means of reducing both false positives and false negatives.