A Nonphysics-based Approach for Vibration-based Structural Health Monitoring under Changing Environmental Conditions

In this article, a technique is proposed to detect damage in structures from measurements taken under different environmental and operational conditions. The method is based on the regression analysis of the features extracted from the vibration measurement. Macro strain was chosen as the selected feature to be measured using the long-gage distributed fiber Bragg grating (FBG) sensors. Features extracted from the measurements on an intact structure were used to construct a reference model for damage identification. Damage was identified by comparing the slope of the regression line of the subsequent measurements to its counterpart of the reference model. Experimental results show that the extracted features were very consistent. The proposed method was demonstrated and validated using noise-polluted numerical simulation data from two different types of structures, a bridge girder and a plane frame, as well as experimental results from a steel beam with different damage scenarios under changing environmental conditions. Different levels of damage were easily identified from the change in slope of the regression lines. The proposed technique is also applicable to statically measured data.