【作者】 王铁生；

【导师】 华锡生；

【作者基本信息】 河海大学 ， 水工结构工程， 2003， 博士

【摘要】 本文以地下隧洞为对象，系统地研究了地下隧洞地面和地下现代测控理论、技术以及各项测控环节的精度分析，盾构掘进导向系统以及施工引起的地表变形动态预测模型和方法。具体研究内容如下： (1)研究了遂洞GPS测控网的布设、数据处理理论；探讨和研究了构筑物反射等产生的多路径效应的影响及应用小波分析理论有效减弱GPS接收信号的多路径噪声的方法；讨论了地下隧洞贯通的误差来源及贯通误差的优化配赋方法。 (2)对竖井联系及地下测控技术和可靠性进行了研究。详细分析了联系测量各因素对方位传递的影响，探讨并比较了加测陀螺方位角后地下导线的精度提高问题，论证了加测陀螺方位角后直伸等边导线终点精度的严密计算公式及陀螺方位边布测的最佳位置。 (3)精密导向系统的研究。盾构导向系统是盾构掘进时的指挥系统，对指导盾构掘进、隧洞贯通、减少地表沉降以及保证隧洞的质量等具有重要的作用。本文在研究盾构导向工作原理的基础上开发研制了一套盾构导向系统，精度高、工作可靠，可极大地缩短导向测量的时间和减轻劳动强度，实现了盾构掘进的实时定位和姿态监测。从理论上分析了各项因素对盾构位置和姿态的误差影响，导出了理论公式，总结了提高精度的相应措施。应用表明其精度较高，而且操作简单，在实际定位和导向中具有很好的应用价值。 (4)研究了隧洞施工中地表沉降动态预测模型，提出了时变参数灰色—时序动态预测模型，并建立了一种改进的时变灰色模型。为了充分利用有限的地表变形数据所蕴涵的内在规律性，提出了利用变形数据的正逆时间序列建立AR模型的方法，并与时变灰色模型组合，不但可反应出变形数据序列的趋势性，同时还可表现出其随机性，从而可进一步提高预测的精度和效果。模型应用于遂洞地表变形预测，验证了模型的有效性和准确性。 (5)探讨和研究了神经网络模型在盾构掘进时引起的地表沉降预测方法及安全性评价。针对BP算法存在的问题，将遗传算法应用于神经网络，并将改进的神经网络模型及模糊神经网络用于盾构推进时地表变形量和变形因素之间的非线性映射关系的建立，进而应用于地表沉降的变形预测和预报。实例表明此模型和方法的有效性和准确性。更多还原

【Abstract】 In this dissertation, the modem monitoring technology based on the advanced theories on underground tunnel is provided; the precision control methods and measures are systematically discussed; the direction guidance system of shield tunneling and the dynamic forecasting models of ground settlement are introduced. The main contents are as follows:(1) The measurement data processing theory and the design scheme of GPS tunnel control network are introduced; A wavelet analysis technique that can efficiently reduces the multi-path effects caused by tall buildings is presented; Finally, the factor sources of transfixions errors of underground tunnel and limit error distribution method which depends on unequal influence principle are proposed.(2) The technology, accuracy, reliability of underground monitoring and control surveying are discussed. The influence factors and optimal scheme of connected-triangle surveying with shaft is analyzed in detail. The formula for calculating the lateral transfixion errors of underground traverse with added gyro orientation line is derived strictly, the calculation results show that the accuracy can be increased obviously by the way of orienting added gyro lines. The optimal position of added a gyro orientation line and increment rate of accuracy is provided.(3) The precise direction guidance system of shield tunneling is studied. The direction guidance system plays an important part in the shield excavation, such as the prevention of snake motion, the reduction of ground settlement of tunnel construction and the insurance of tunnel quality etc. Based on the principle of guidance system, a precise direction guidance system with high accuracy, high efficiency and more reliability is developed, it has the advantages of less times of surveying, real time positioning and shield behaviors monitoring. The accuracy estimation methods of the guidance system are discussed in detail. The practical application in underground tunnel shows the system availability and great practical value.(4) The dynamic data forecasting model of ground settlement is studied, a new prediction model of grey-time serial with time-varying parameters characters is proposed. Based on the analysis of the shortcoming of the classic grey model, an improved grey model with time varying parameters is presented. An improved AR model is studied, which established by the combination normal order time serial and contrary order data in case the observations are less, and then, the combination model with improved grey and time serial is introduced. It can reflect not only the deformation tendency, but also the stochastic characters. It is very suitable to be applied to deformation analysis and prediction. Finally, the effectiveness and accuracy of ground settlement predicting in a practical underground construction with the combination model are validated.(5) The NN(Neural Network) prediction model of ground settlement in the shield tunneling is proposed. Firstly, the shortcomings of the back propagation algorithm, such as slow learning speed and local minimum points, are discussed. Then an improved hybrid model with fast convergence rate, good performance based on combination of the genetic algorithm and the BP algorithm is presented by improving real coding, genetic operators etc. Finally, the prediction models of NN and FNN(Fuzzy Neural Network) are tested in shield tunneling, and the experimental results show the model has the higher forecasting accuracy. Then the conclusion is drawn that this method is an effective way of solving prediction problem of tunnel construction.更多还原