Abstract
AbstractThis paper proposes a hybrid crack estimation technique that utilizes digital images from cameras and physics-based simulations to perform online diagnosis and prognosis of miter gate. To fully capture the localized effect of the crack, a global-local coupled finite element (FE) model is first created. An iterative global-local (IGL) algorithm is then developed to provide increased accuracy over sub-modelling at the expense of increased computational cost. To replace the process of solving the complex local FE, a Gaussian process (GP) surrogate model is further constructed to increase the computational efficiency. By interpolating the nodal displacement values collected from the surface around the crack, another GP surrogate model is developed to generate synthetic images similar to that obtained from cameras. The results demonstrate that the proposed method can efficiently predict the parameters of the crack growth model as well as to estimate the true crack length.KeywordsMiter gatesGlobal-local iterationSurrogate modelBayesian networkStructural health monitoring
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
