Time Varying Analysis Model for Cracks Causality of Concrete Dams

The main factors of influencing the crack change of concrete dams can be reduced to three parts of water pressure, temperature and aging, but it is very difficult to get the effects of this three parts respectively. The nonlinear time-varying relationship between the crack change and its influencing factors is very strong. Using the unique ability of data analysis and learning of artificial neural network, and through the intercrossing of dam construction knowledge, then a good approach to solve these problems can be offered. In this paper, based on in-situ monitoring data of concrete dams, from the angle of machine learning, the time varying analysis model for causality of cracks of concrete dams is introduced, and the evolution law of load cracks with the change of time and load is analyzed and forecasted. The example indicates that the model can finely describe and forecast the time varying characteristic of cracks causality in concrete dams, and can give each factor's contribution to crack opening.