关键词: 隧道大变形, 单项预测, 组合预测, 变形趋势判断, M-K检验

Abstract: In order to improve the prediction accuracy of tunnel large deformation, ARMA model, cubic exponential smoothing model and GM (1,1) model were firstly used as single prediction model to predict deformation respectively. Then a variety of linear and non-linear combination models were used to realize the preliminary combination prediction. On the basis of progressive combination prediction and chaotic optimization prediction, an integrated combination prediction model for large deformation of tunnels was established to realize the comprehensive prediction. Meanwhile, M-K test was proposed to judge the development trend of tunnel large deformation, so as to verify the accuracy of prediction results. Engineering prediction results showed that all kinds of preliminary combined prediction models can improve the forecasting accuracy in varying degrees, and progressive combined prediction model can optimize the residual sequence and achieve the purpose of improving the forecasting accuracy, that is, the integrated prediction model has a high forecasting accuracy in tunnel large deformation prediction. At the same time, the results of deformation prediction are consistent with those of trend judgment. It is concluded that the large deformation of the tunnel tends to develop in an unstable direction, which corroborates the accuracy of the analysis results, and provides a reference for disaster prevention and control.

Key words: tunnel large deformation, single prediction, combined prediction, deformation trend judgement, M-K test