Study On Horizontal Displacement Prediction Of Retaining Structure Of Deep Foundation Pit Based On Generalized Regression Neural Network

The21st century will be the era of underground space developmentunprecedented prosperity. Underground space and the construction of high-risebuildings are inevitably involve deep excavation. Excavation system and thesurrounding soil is highly complex space systems composed by a variety of media,foundation deformation and safety are subject to common variety of factors. So theinformation is necessary to carry out dynamic construction, which requires theconstruction process monitoring to be performed in a timely manner.Through themonitoring of data collection, analysis, forecasting and feedback, It can adjust theparameters of construction and engineering measures.The traditional predictiontheories of pit deformation exist different degrees of limitations.Therefore, proposingapplicability of the theory and methods of forecasting pit deformation has importanttheoretical and practical value.In this paper, basing on an in-depth analysis of characteristics of horizontaldisplacement prediction of deep foundation pit, GRNN(Generalized RegressionNeural Network) predicts horizontal displacement of deep foundation pit. UsingMATLAB neural network toolbox to analyze the GRNN’s structure,optimization ofinput sample size, optimization of parameters.The GRNN compares with GM (1,1)and time series AR model to predict the results of comparative analysis.It shows thatGRNN model to predict the horizontal deformation of deep foundation pit is feasiblewhich meets the forecast accuracy. Further exploration on prediction methods ofhorizontal displacement of retaining structure of deep foundation pit based onhistorical displacement, multi-source data (excavation depth and axial force) and thecombination of the fommer both factors.