Study On Self-Memory Prediction Model Of Foundation Displacement Time Series Analysis

In the geotechnical engineering, foundation pit engineering is a complex existence for whose deformation is integrated influenced by many factors. The traditional designing methods for controlling deformation are difficult to achieve the desired effects. So method that to master the deformation condition of supporting structure and surrounding environment through site construction monitoring and then to analysis and process the monitoring data for predicting the deformation trend and promptly optimal designing and guiding the construction is the main method to avoid accidents happened. To improve the accuracy and stability of foundation displacement prediction have great significance for researching the new method for predicting foundation displacement.Pit displacement system has its certainty and randomness. While the traditional time series analysis method of foundation displacement is commonly based on stochastic theory which belongs to an uncertainty method. This paper tries to introduce a nonlinear method namely dynamic system self-memorization principle which is combined with random and dynamic into the pit displacement prediction, and to seek a new way for predicting displacement. The paper is down below:Firstly, studying the method that using traditional two-way finite difference method to inverse dynamic differential equation of displacement time sequence and then building self-memory model of two-way difference on this basis.Secondly, utilizing the grey system theory to inverse gray differential equation of displacement dynamics system and taking it as the dynamic differential equation to build grey self-memory modelThirdly, proposing a method that to seek system dynamic differential equation through the curve fitting method based on the existing research and then establishing trend curve-self-memory forecasting model.We apply three self-memory models in predicting the horizontal displacement prediction of Foundation Pit supporting structure in Shenzhen metro shopping park station. The results show that all three self-memory models have good fitting and predictive accuracies. But the derivation processes of two-way difference self-memory model are complicated, tedious calculation and heavy workload for modeling. However, grey self-memory model and trend curve self-memory model have better maneuverability than two-way difference self-memory model. Moreover, the trend curve self-memorial model has better fitting and forecasting results when volatility points are happened to time series. So it is a more simple and practical self-memory forecasting model.This article introduces self-memorization principle into the time series analysis of foundation pit displacement and for which it provides a new theory and new method in foundation pit displacement prediction.