Abstract In order to precisely predict the deformation of a tunnel's surrounding rocks, the principles of radial basis function neural network and non-equal time-interval GM (1,1) grey system theory were adopted to establish prediction models. The fitting programs for two prediction models were compiled by way of the MATLAB-2010b platform. Based on the measured rock deformation data from the Linli tunnel on the Zhangzhuo Highway, the deformation prediction was carried out by model training. According to a comparative analysis of the predicted values and measured values and a model error verification test, it was revealed that the prediction results of a radial basis function neural network (RBFNN) model are more precise and that the predicted deformation curve agrees better with that of measured one, truly reflecting the variation law of surrounding rock deformation.
Abstract：
In order to precisely predict the deformation of a tunnel's surrounding rocks, the principles of radial basis function neural network and non-equal time-interval GM (1,1) grey system theory were adopted to establish prediction models. The fitting programs for two prediction models were compiled by way of the MATLAB-2010b platform. Based on the measured rock deformation data from the Linli tunnel on the Zhangzhuo Highway, the deformation prediction was carried out by model training. According to a comparative analysis of the predicted values and measured values and a model error verification test, it was revealed that the prediction results of a radial basis function neural network (RBFNN) model are more precise and that the predicted deformation curve agrees better with that of measured one, truly reflecting the variation law of surrounding rock deformation.
2015, Vol. 52 