A prediction-error-based method for data transmission and damage detection in wireless sensor networks for structural health monitoring

In this work, a prediction-error-based method that aims to reduce the amount of data transmission and storage in structural health monitoring wireless sensor networks (WSN) while keeping the damage detection capabilities intact is presented. WSNs offer tremendous promise for accurate and continuous structural monitoring using a dense array of inexpensive sensors and possess many advantages over conventional wired systems, particularly for large civil infrastructures. This paper validates the studied approach with the data collected during a test on a three-story steel frame mounted at the ELSA laboratory of the EU Joint Research Center in Ispra, Italy. The frame is tested with and without steel bracing, which results in non-sway behavior and sway frame responses. The excitation is provided by a shaker mounted on top of the frame. Accelerometer signals are collected by a multi-channel data acquisition system. The proposed method has been tested extensively via recorded experimental data and it offers considerable saving in transmitted energy while at the same time keeping the ability to detect changes or damage in the structure.