Study On Deep Foundation Pit Deformation Monitoring And Deformation Prediction

This paper mainly studies the deformation monitoring and prediction in the process of deep foundation pit engineering construction.It takes Wanda Plaza deep excavation in Jixi City, Heilongjiang Province as the background,summarizes the deep foundation pit monitoring technology in an all-round way,analyzes the GM(1, 1) model, neural network model and combination model and its forecast effect and verifies its reliability and practicability.The main contents and conclusions are as follows:1.In the paper,the mechanism of the deep foundation pit deformation and deformation monitoring technology are introduced,the layout of deformation monitoring system and mo nitoring data processing and so on are discussed.The paper introduces the basic theory of grey GM(1, 1) model, BP neural network model and Elman neural network model, analyzes the application of the deep foundation pit deformation prediction of series combination of grey model and neural network model.2.The grey GM(1,1) model can be qualified for deformation monitoring data fitting and forecasting,it is unfavorable to make the the forecast period too long.3.The designed BP and Elman neural network can be applied to the deep foundation pit deformation forecast practice after the training,also can achieve higher fitting precision,and fully meet the requirements of practical engineering application.Basing on the research and analysis of the resultsof the sample data,the prediction precision of BP neural network model is highest.It’s a close prediction accuracy of Elman neural network and BP neural network.The Elman neural network does not show the superiority of forecasting accuracy compared with its advantages of appling to other forecasting aspects. The BP neural network should be more suitable for application in deformation prediction.4. Combining with the study of concrete examples, the prediction accuracy of combined model is higher than that of single GM(1, 1) model.The combined model can live up to their respective strengths,both can use the highly nonlinearity of the neural network and can make use of accumulating data regularity and can take advantage of the randomness of Weakening data.5.By comparing the results of several kinds of models in this paper, it can be concluded that the GM(1, 1) model, neural network model, the combination model can predict accurate results and can effectively guide the construction of foundation pit engineering.