Vibration response based crack diagnosis in beam-like structures inference

Cracks in Beam-like structures can cause unexpected destruction of the structure leading to unintended damage to property and human life. This damage can be prevented by early detection of the crack. In this work, a Fuzzy Inference System is used to predict crack depth and location in a beam. The vibration signature of a healthy (uncrack) beam and a set of crack beams were obtained experimentally and characterized by the first Natural Frequency and statistical kurtosis. The inputs to the Fuzzy system were the first Natural Frequency and kurtosis of the vibration response signal. A system of Fuzzy rules was established by machine learning from the experimental data sets and applied to Triangular, Gaussian, Trapezoidal and Bell-shaped membership function sets. Dispersed standard deviations around the mean absolute errors were observed for the predictions. It was concluded that more clustered data sets for establishing the Fuzzy rules could improve precision of crack diagnosis. The presented study is capable of being used in an online Structural Health Monitoring algorithm to identify crack depth and location. (C) 2021 The Author(s). Published by Elsevier B.V. on behalf of African Institute of Mathematical Sciences / Next Einstein Initiative.