管幕工程地表变形的人工神经网络预测研究

管幕工法是一种新型的地下空间暗挖技术.国内首次应用于上海市中环线浦西段的北虹路下立交工程.该工法为解决软土地区超大断面地下工程的施工变形问题开创了新的领域.管幕工法虽然可以减少对周边环境的影响,但并不能完全消除.作者在充分研究钢管幕顶进施工过程中引起的地表变形特征基础上,建立相应的预测模型,应用人工神经网络智能滚动预测方法,对管幕工程的地表变形进行预测研究.研究表明：人工神经网络的一步滚动预测可以满足实际的工程需要,但精度相对偏低.而多步滚动预测虽可以得到较高的预测精度,但在实际工程应用中还需解决量的优化问题.

Intelligent Prediction Study of Ground Displacement in Pipe Roofing Engineering by Artificial Neural Networks(ANN)

A new type of hidden excavation method the Pipe-roofing one has been firstly employed in China in the Beihong Road underground intersection project of Puxi segment of Shanghai Middle Ring Road and this creates a new field for the displacement controlling problem of large cross-section tunnel in soft soil area.Although this method can reduce the effect of excavation on surrounding environment, there still is displacement that should be concerned about.In this paper,the intelligent stepping displacement prediction model for Pipe-roofing method has been erected based on Artificial Neural Networks theory after serious studies on the ground displacement characteristics of Pipe-roofing method and its prediction experience in the Beihong Road project has proved that Single-step stepping method can give prediction result to meet the need of the project,but it is hoped to be more accurate.While Multi-step stepping method can give result with high precision,but it is not convenient for application because of the burdensome prediction process.There are still optimizing works to be done in the near future.