Wireless sensor network-based pattern matching technique for the circumvention of environmental and stimuli-related variability in structural health monitoring

Much research has gone into using wireless sensor networks to monitor structural health by sensing and measuring vibrations. One problem here is that structural vibrations can be affected by many factors, which can make it difficult to determine the contribution of structural condition to measured vibrations. We propose a robust solution to this difficulty that consists of a wireless sensor network that implements a highly efficient, fully distributed pattern matching algorithm. Here, we exploit correlation between sensed vibration signals at different locations on the structure under measurement to detect damage. Potential applications for our system are numerous. They include many infrastructure applications such as those involving railroad and pipeline monitoring. The general solution is described, including tradeoffs between accuracy and energy/memory consumption. It is shown that the accuracy of our approach grows gracefully at the expense of memory and energy consumption. Additionally, a case study involving railroad applications is discussed. Simulations in our case study indicate that our distributed approach can reduce the consumed energy for transmitted data by 50% compared to a centralized architecture.