复合地层小直径隧道掘进机掘进速度区间预测

合理准确预测隧道掘进机(tunnel boring machine,TBM)的掘进速度是实现TBM智能化控制的关键问题之一,复合地层小直径TBM施工的不确定性较常规地质条件更强,而传统预测方法对施工过程的不确定性考虑不足。在此通过引入区间预测方法,提出基于4种不同Bootstrap方法结合KELM-ANN模型的TBM掘进速度区间预测模型,并以南水北调安阳输水隧洞工程为例,选取142组工程实测数据验证区间预测模型的有效性。研究结果表明:基于Rademacher分布建立的模型预测结果优于其他3种方法,不仅可以得到较好的点预测结果,还可以构造出较为清晰可靠的区间将掘进速度实测值完全包络在内;随着置信水平的提高,区间可容纳的不确定性和风险也逐渐上升,通过变化区间宽度,能较好地量化和解释TBM施工过程中的不确定性因素对掘进速度的影响。研究结果可为TBM掘进性能预测和掘进参数优化提供参考。

Study on the interval prediction model of small-diameter TBM penetration rate in mixed face ground

Abstract] Reasonable and accurate prediction of TBM tunneling penetration rate is one of the key issues to realize TBM intelligent control. The uncertainty of small-diameter TBM construction in mixed face ground is stronger than conventional geological conditions, and traditional prediction methods do not consider the uncertainty of the construction process. In this paper, the interval prediction method is introduced, and an interval prediction model of TBM penetration rate based on 4 different Bootstrap methods combined with the KELM-ANN model is proposed. Taking the Anyang water conveyance tunnel project of the South-to-North Water Diversion Project as an example, 142 sets of measured data are selected to verify the effectiveness of the interval prediction model. The research results show that the prediction results of the model based on the Rademacher distribution are better than those of the other three methods. Not only can better point prediction results be obtained, but also a clearer and more reliable interval can be constructed to completely enclose the measured value of the penetration rate; With the improvement of the confidence level, the uncertainty and risk that can be accommodated in the interval also gradually increases. By changing the width of the interval, the impact of the uncertainty factors in the TBM construction process on the penetration rate can be better quantified and explained. The research results can provide reference for TBM tunnel performance prediction and tunnel parameter optimization.