Estimating mechanical rock mass parameters relating to the Three Gorges Project permanent shiplock using an intelligent displacement back analysis method

Establishing the mechanical rock mass parameters is one of the important tasks for the highwall stability analysis of the permanent shiplock at the Three Gorges Project in China. Existing back analysis methods are not sufficient to provide the necessary accuracy and to recognize non-linear relations. The new displacement back analysis method proposed in this paper is a combination of a neural network, an evolutionary calculation, and numerical analysis techniques. The non-linear relation involving displacement and mechanical parameters is adequately recognized by the neural network techniques. The neural networks learn using an evolutionary technique, with samples created by orthogonal design and tested with new cases given by event design. With the neural network model established, the mechanical parameters are recognized using a genetic algorithm over a large search space in the global range. The predicted displacement occurring for each excavation step from January 1998 to the end of excavation and their cumulative values for 5 later excavation steps are closely characterized by the new analysis technique.