Comparative Study On The Monitoring And Prediction Method Of Settlement Deformation Of Soft Soil Subgrade Of Zhongshan Ring Expressway

In order to discuss the settlement law and characteristics of soft soil subgrade under different treatment methods,and to study its future settlement development trend.Based on the measured data of the settlement monitoring of the soft soil subgrade of the Zhongshan Ring Expressway,this paper analyzes the development law of the settlement of the soft soil subgrade under different treatment methods.Four prediction methods are used to fit the measured settlement curve,and the settlement development trend of soft soil subgrade is predicted and compared with the measured value,and the method with the highest prediction accuracy is compared.At the same time,according to the parameters of the geotechnical test and the soil in-situ test,the finite element models of the soft soil subgrade with two different foundation treatment methods,the natural soft soil subgrade and the pile-slab treatment,were established to simulate the actual engineering.The main research results are as follows:(1)Study the physical and mechanical properties of soft soil and the consolidation and settlement mechanism of soft soil subgrade,and analyze the reasons for the adverse effects of soft soil subgrade on the project.(2)According to the monitoring data of the settlement plate,analyze the variation law of the settlement amount of the soft soil subgrade with time,and analyze the variation law of the horizontal displacement of the soft soil subgrade with time according to the displacement pile and the inclinometer pipe.The deformation characteristics of the soft soil subgrade with different treatment methods under the piled load of the gradual filling are analyzed.(3)In order to better understand the development trend of soft soil subgrade settlement and avoid possible future risks.The traditional curve fitting method,combined prediction model and BP neural network improved by genetic algorithm are used to predict settlement.The order of prediction accuracy of each prediction method for soft soil subgrade with different treatment methods is:genetic algorithm improved BP neural network>combined prediction method(optimized combination of two traditional curve fitting methods)>traditional curve fitting method.Among them,the genetic algorithm improved BP neural network has high prediction accuracy of the soft foundation settlement in this area,and can accurately control the future development trend of the soft foundation settlement.The combined forecasting model can improve the overall forecasting effect of the model by synthesizing the systematic quantities of the two single forecasting models,but it can only be used to combine two forecasting methods with similar forecasting accuracy.(4)In order to further study the long-term settlement change law and consolidation mechanism of soft soil subgrade,the ABAQUS finite element software was used to calculate the settlement and stress-strain analysis of soft soil subgrade in the process of piling preloading and 15 years after the completion of the pavement under different treatment methods.It is concluded that the soft foundation of the general road section is suitable for treatment using plastic drainage boards,and the soft foundation at the culvert,bridgehead section and passage is suitable for treatment using cement mixing piles.