Abstract:

The actual geometric size of dam foundation is uncertain. In this research, a neural network model for the intelligent identification of the uncertain geometric size of dam foundation is established. The model takes the relative displacement of monitoring points as input, and the dam concrete, rock foundation material parameters and foundation’s geometric size as output. The load distribution of steady seepage body is obtained, and on the basis of material parameters combined according to uniform design principle, the relative displacement of key monitoring points were calculated as the learning samples. The trained network describes the nonlinear relationship among the dam concrete, rock foundation material parameters and the foundation’s geometric size and dam deformation. The water pressure component separated from the measured dam displacement is input into the trained network to automatically identify the dam concrete and rock foundation material parameters and the foundation’s geometric size. Calculation example shows that this model is feasible.

Key words: foundation’s geometric size, uncertainty, intelligent identification, concrete dam