Research On Intelligent Back Analysis Of Tunnel Surrounding Rock Displacement And Engineering Practice

Now there are some shortcomings in the methods to gain the parameters of tunnel surrounding rock. It is believed that displacement back analysis method can solve the problem, but there are still some defects in it, for example, not solving the optimization problem, and the methods is inefficient and are so difficult to be used that are not feasible in engineering.Aiming at the above problems, the dissertation founds a network of intelligent back analysis of tunnel surrounding rock displacement which is based on ANN(artificial neural network) toolbox of matlab, according to road tunnels and on-site measured data. The network is based on BP artificial neural network which has powerful nonlinear functions mapping abilities, auto learning and generalization ability. Bayesian regularization method is used to improve the network's generalization ability. In order to ensure that the network's training samples are in accordance with reality and have uniform dispersivity and symmetrical comparability, the dissertation simulates the surrounding rock displacement of excavating tunnel with early lining by fast lagrangian analysis of continua program. The results of simulation accord with fact, then make experiments with orthogonal design and gain training samples. The result proves that the precision of back analysis can reach above 90%.At last, an engineer example validates the network's applicability. When it is applied to the practical engineering, a certain economic performance is obtained.