Settlement Prediction Of Soft Soil Roadbed Treated By Cfg Pile

With the rapid development of urban economy in China,great achievements have been made in urban road construction,but the settlement of soft soil roadbed has been plagued by construction.Due to geographical location and environmental constraints,many new projects in Zhaoqing City,Guangdong Province have to be built on soft soil foundation.In order to ensure the long-term stability of subgrade,soft soil foundation must be treated.Accurate prediction of roadbed settlement is of great significance for controlling construction qualityBased on the project of "New Construction of Qiaobei Road,Duanzhou District,Zhaoqing City",this paper studies the treatment of soft soil foundation with cement fly ash gravel pile(CFG pile),and predicts the settlement of the treated foundation.The achievements of this paper are as follows:(1)By participating in the actual project construction,the construction method and key points of CFG pile in soft foundation treatment of Zhaoqing municipal road are studied.At the same time,the settlement of treated foundation is monitored by monitoring instruments,and the monitoring data are analyzed to predict the development trend of settlement and to study the effect of CFG pile in soft foundation treatment.(2)Comparing the precision of hyperbolic fitting method and exponential curve fitting method to predict settlement,the final settlement prediction value and settlement fitting curve are obtained.(3)On the basis of BP neural network,two improved methods,BP algorithm with momentum term optimization and momentum BP algorithm with adjusting learning rate optimization,are adopted to realize network+optimization.The settlement value predicted is very close to the measured value,which proves that it is feasible to use this method to predict roadbed settlement.The settlement curve obtained by the momentum BP algorithm which adjusts the learning rate is in good agreement with the measured settlement value.By comparing with curve fitting method,it is found that the dynamic prediction method of BP neural network is better than the static prediction method.