Date Added: Jun 2010
This paper presents a localized information processing approach for long-term, online Structural Health Monitoring (SHM) using Wireless Sensor Networks (WSNs). Based on the embedded AR-ARX method, each sensor independently calculates a statistical damage-sensitive coefficient using the measured acceleration data during each monitoring period. A non-linear programming formulation is developed to identify damage presence, localize damage position, and quantify damage severity from the damage-sensitive coefficients in the whole sensing field. By limiting each sensor to exchange information among its neighboring sensors only, a localized near-optimal algorithm is proposed to reduce communication costs, thus alleviating the channel interference and prolonging the network lifetime.