关键词: 软岩隧道, 大变形, LS-SVM, 预测模型, 灰色模型, BP神经网络

Abstract:

In order to realize the judgment of the development trend of large deformation of tunnel, optimize the site construction, and avoid construction safety risk, the LSSVM and optimized GM (1, 1) models are adopted to predict the deformation of tunnel. And then the quadratic sum of error is taken as an index, and the prediction results of the two models are combined. Finally, the BP neural network is used to modify the prediction error so as to realize comprehensive prediction. The case test results show that the optimization effect on support vector machine by least squares method is superior to that by grey model; and the error modification model can further effectively improve the prediction accuracy, which make the predicted and measured values closer. Meanwhile, the prediction results show that the deformation of later 4 cycles are still with continuous deformation trend, which indicates that effective measures should be adopted to avoid accidents. The abovementioned prediction model has better prediction accuracy and applicability, which has a certain significance for the study of large deformation of tunnel.

Key words: soft rock tunnel, large deformation, LSSVM, prediction model, grey model, BP neural network