Damage localization using a power-efficient distributed on-board signal processingalgorithm in a wireless sensor network

Abstract

A distributed on-board algorithm that is embedded and executed within a group ofwireless sensors to locate structural damages in isotropic plates is presented. Thealgorithm is based on an energy-decay model of Lamb waves and singular valuedecomposition (SVD) to determine damage locations. A sensor group consists ofa small number of sensors, each of which independently collects wave signalsand evaluates wave energy upon an external triggering signal sent from a basestation. The energy values, usually a few bytes in length, are then sent to the basestation to determine the presence and location of damages. In comparison withtraditional centralized approaches in which whole datasets are required to betransmitted, the proposed algorithm yields much less wireless communication traffic, yetwith a modest amount of computation required within sensors. Experiments haveshown that the algorithm is robust to locate damage for isotropic plate structuresand is very power efficient, with more than an order-of-magnitude power saving.