This paper presents a novel fatigue crack monitoring method for steel specimens based on the smoothness priors method (SPM) and Tsallis entropy (TE) of strain measurements. The aim of the study is to detect initiation of a crack in steel specimens and subsequently to monitor its propagation under the fatigue load, based on real-time strain measurements. The nonlinear dynamic response of the structure was exploited since it degrades due to the initiation and subsequent propagation of the crack under the external dynamic excitation. The proposed method was experimentally validated. Here, the SPM is applied to decomposing the structural strain response into a nearly-stationary (NS) component and a low frequency aperiodic trend (LFAT) component. Features associated with crack initiation can be extracted from the NS component. The LFAT component, on the other hand, can be used to identify crack propagation. To tackle the singularity of the structural responses associated with a crack, the TE of the NS component was used in detection and monitoring of the crack in the steel specimen. Two other techniques, namely, acoustic emission (AE) sensor and crack opening displacement (COD) gauge were used for the purpose of calibration and comparison. The results show remarkable resemblance in terms of crack initiation and propagation identification exhibited by all three types of sensors, highlighting the potential of the proposed method for real-time detection and subsequent monitoring of crack propagation in steel structures.
Experimental validation of textured sensing skin for fatigue crack monitoring