Assistant Professor, Department of Civil Engineering, Roudehen Branch, Islamic Azad University, Roudehen, Iran
Abstract
Although structural health monitoring systems have considerably advanced during the last decades, they are still expensive. Moreover, the existing low-cost systems use numerous sensors. Therefore, due high price of sensor hardware they are uncommon in our country. Owing to this, reduction of sensor numbers can be a step forward toward naturalization of such systems. In this paper, a health monitoring algorithm is proposed base on signal processing. Only one sensor is employed to detect damage. Since reduction of sensor numbers can affect the accuracy of damage detection, in the first step of the paper, several signal processing techniques are compared and the best one for this case is selected. Next, two damage indices are introduced based on frequency dispersions. Time-series analysis is utilized to tackle fluctuation of damage indices. Finally, a damage threshold is defined to distinguish damaged and healthy states. Results show that frequency itself is not an appropriate damage measure. Instead, frequency dispersion can be used as an alternative damage criteria.