Study On Monitoring Data Analysis And Prediction Of Foundation Pit Engineering Based On Chaos Theory

Engineering monitoring has become a mandatory requirement of the constructionand management department in geotechnical engineering, especially in the deepfoundation engineering. The monitoring data reflect the real situation of all aspects in theconstruction process. It has very important engineering significance, scientific researchsignificance and economic significance. Therefore, the monitoring data analysis usingdifferent theoretical methods have been gradually become one of the important researchtopics in the field of geotechnical engineering.This thesis based on the monitoring data of a station’s foundation pit in Wuhan. Andthen, the author use chaos theory and the programming tool matlab7.1to analyze themonitoring data and make deformation prediction. The monitoring data come from threemajor monitoring projects: the maximum horizontal displacement of deep soil, buildingsettlement beside the pit and the steel brace stress. First, I calculated the delay time andembedding dimension of the monitoring data using C-C method. Then, calculated thelargest Lyapunov exponent and the correlation dimension, which are the most importantchaotic characteristics, using the method of small data. And then, analyzed itsmathematical meaning of the chaotic characteristics quantitatively, and analyzed itschaos performance in the excavation process qualitatively; At last, created two chaoticmodels: the one-order weighted local prediction model and the largest Lyapunovexponent prediction model, to predict the deformation of each measuring point.Through the analysis of the contents above, the main conclusion could be obtainedin this thesis as follows: The change of the steel brace stress is more complex andsensitive than the displacement of the deep soil and the settlement of the construction.Because the delay time and embedding dimension of the time series of the steel brace arelarger than others; The correlation dimension of each time series is a non-integer. Itmeans that these points all have chaotic characteristics. The largest Lyapunov exponentof most points is positive. And average cycles have a greater impact on the largestLyapunov exponent; The one-order weighted local prediction model has good predictiveability in various points. The relative error of predictive value of the building settlementis less than2%. On the other hand, the model based on largest Lyapunov exponent showed instability in the forecasting process; As can be seen from the predictionaccuracy of two forecasting methods for each measuring point, the one-order weightedlocal prediction model is better than the other model in this excavation engineering.