| With the rapid development of water resources and hydropower construction in China, the seismic resistance study on high earth-rock dams is very important. As one of the essential contents of the study, it is a necessary to find a reasonable calculation model for the analysis of earthquake-induced permanent deformation and find a way to obtain relevant parameters accurately. It is beneficial to predict the dynamic response, and then provide a basis for not only earthquake-resistant design but also safety and stability evaluation. In order to solve the problem, two parts of the main work in this paper are as follows:1. Radial basis function neural network(RBFN) was used to back-analysis dynamic parameters of earth-rock dams, a kind of parameter inversion model for earthquake-induced permanent deformation was proposed based on RBFN, and the model was applied to have a try on Zipingpu CFRD. The result shows that the proposed model is efficient in back-analysis of earthquake-induced permanent deformation parameters of earth-rock dams. And the existing calculation theories and methods can basically reflect the earthquake-resistant characteristics of earth-rock dams, but it’s still necessary to study the dynamic analysis method further, in order to make permanent deformation parameters inversion more meaningful.2. To further improve the accuracy of parameters inversion, the effects of incident direction of seismic waves and reservoir water level on seismic response of the dam were focused on, taking Zipingpu CFRD as an example. The result indicates that the maximum responses of acceleration and dynamic displacement are affected intensely by the incident direction of seismic waves, and they apparently changed especially when the angle between incident wave and dam axial varies from-15°to15°And that the natural frequency of the dam decreases with the increase of water level, the variation rate is slow first and then faster. Besides, the main impact of water level is on the acceleration of dam axial, the other seismic responses, such as dynamic displacement and permanent deformation, have relatively obvious change only when the water level is as high as4/5height of the dam. As a consequence, when the analysis of dynamic response and relevant parameters inversion is performed, the water level should be taken into consideration if the water level up to the corresponding height.In this work, RBFN is extended to the field of dynamic parameter inversion for earth-rock dams, which solved the previous problem that dynamic parameters inversion was time consuming. And the research on the effects of incident direction of seismic waves and the reservoir water level is going to further improve the accuracy of parameter inversion. |
