基于MIV-BP模型和AIC准则的盾构掘进参数优化研究

为解决水下隧道和隧洞工程盾构施工关键掘进参数在复合地层中难以控制的问题,以广东陆丰核电站1、2号机组排水隧洞工程盾构施工为背景,将平均影响值(Mean Impact Value,MIV)算法引入BP神经网络模型,筛选出对施工效果影响显著的关键掘进参数,在此基础上,基于AIC准则对其进行最优分布拟合,提出以50%和90%置信水平下的置信区间,分别作为掘进参数的控制区间和预警区间的掘进参数优化设计方案,并基于Python脚本语言自带的开源Scikit-Learn、SciPy模块库开发了相应的程序。分析结果表明,刀盘扭矩、总推力等参数对隧洞拱顶沉降起重要的控制作用,所开发程序具有良好的统计分析、快速指导施工的功能,可以为同类型盾构在相似复合地层下掘进参数的选取、优化和隧洞拱顶沉降量的控制提供参考。

Research on Boring Parameters Optimization for Shield Based on MIV-BP Model and AIC Criterion

Aiming at the problems that the key boring parameters in mixed ground are more difficult to control in submarine drainage tunnels and caverns, this paper proposes a specific solution. In this study, based on the engineering application example of submarine drainage tunnel project of No.1 and No.2 Unit in Lufeng Nuclear Power Station, the Mean Impact Value(MIV) algorithm is introduced to the BP neural network model and the key boring parameters that have significant impact on shield construction effect can be screened out. On this basis, with the optimum distribution fitting of boring parameters based on Akaike Information Criterion(AIC), this paper proposes an optimized scheme for boring parameters, which use the confidence intervals of 50% and 90% confidence level as the controlling intervals and warning intervals respectively. The corresponding program is developed based on the open source module libraries of Scikit-Learn and SciPy of Python. The analysis results indicate that boring parameters such as cutter head torque and total thrust, have produced significant influences on the control in vault settlement of the tunnel. The developed program has some practical function of statistic analysis and quick guidance of construction, which provide the reference of selection and optimization of boring parameters and the control of vault settlement for the same type of shield construction in similar mixed ground.