Prediction And Control Of Ground Surface Settlement Due To EPB Shield Tunnelling Using Non-linear Method

ABSTRACT:As predicted that the beginning of the21st century to the middle, is the era of China’s large-scale construction of urban subway, this has been proved by practice. In the construction process, Earth Pressure Balance (EPB) shield is the kind of construction method which is often used. In order to control the settlement be accepted. Taking Changsha Metro Line’s construction experiences as example, do research on predicting the settlement and optimizing the EPB’s parameters of construction processing. The main content as follows:Firstly, there has so many researchers has engaged in researching the factors which affects the settlement. This article will summarize the factors from three aspects which including the tunnel geometry factors, formation conditions and the EPB’s construction parameters. Which will be usefull for further predictions and controlling of settlements.Second, taking Changsha Metro Line shield construction interval as background, to analysis the monitored settlement datas, trying to find laws of settlement caused by EPB construction. And to statistic the EPB’s construction parameters, to obtain the laws of parameters range. These experiences will be usefull for other similar projects to predict and control their settlement.Third, to predict and control settlement caused by EPB construction. Taking Back Propagation (BP) Neural Network as tool for prediction, and Genetic Algorithm (GA) as method of optimizing EPB’s tunnelling parameters. Considering the C#as powerfull programming, can form interface easily, and perform efficient. Matlab is also an excellent programming in handling large number of datas and graphics rendering. So, this article combined with these, developed a software which named Neural Network for Settlement Prediction and Controling (NNSPC), which achieved the surface subsidence prediction and control by using the monitor data from Changsha Metro Line. Forth, based on the monitored settlement datas from Changsha Metro Line. Use the written software (NNSPC) as tool for prediction. Training the data until the network results satisfied, building a non-linear relationship between the factors and the settlement. Then, using the neural network for settlement prediction in Changsha Subway Line’s Tujiachong station to Tiedaoxueyuan station which is also constructing by EPB method.Fifth, with the characteristics of Genetic Algorithms (GA), it could find the best individual from a large population. Using the software based on the character, to optimize the EPB’s parameters which the section beneath the Beijing-Guangzhou railway of Changsha Subway Line, And the section beneath the levees of Xiangjiang River of Changsha-Zhuzhou-Xiangtan intercity railway. Give suggestions based on the results for the construction process.