Abstract:

The accuracy of rock mass classification has close relationship with the safety and cost of tunnel construction. Due to the disadvantages of the present rock mass classification methods, a new rock mass classification method based on radial basis function neural network is introduced on basis of the national BQ rock mass classification standard,a large number of site measurements and indoor tests, as well as the rock mass classification practice made in the construction of the tunnel group on NinguoJixi highway. The classification results are served as the training samples for geneticRBF neural network training. Therefore, the intelligent classification model is established after network training has been finished. The application of this model shows that the result of the classification made on basis of the geneticRBF neural network agrees with that made on basis of site reconnaissance. The geneticRBF neural network developed provides a new approach for the classification of surrounding rock mass of tunnels.

Key words: tunnel engineering, surrounding rock mass classification, radial basis function, neural network, genetic algorithm