Ground Pit Subsidence Monitoring Data Processing And Deformation Prediction Model Study

The construction of the city reflects the rapid development of the country.In recent years,the phenomenon of using urban land is increasingly serious.Therefore,the expansion has to be carried out in the air or underground in order to seek space for the development of urban construction.And it accelerates the development of the construction of foundation pits rapidly.On the one hand,the rise of foundation pit engineering has accelerated the construction of cities.On the other hand,safety accidents caused by foundation pits have also occurred frequently.Therefore,the research topic of the monitoring of deformation of foundation pits has drew a lot of attention by scholars.Therefore,mastering the deformation law of foundation pits is an important link to ensure the safety of foundation pit projects and people's lives and property.This paper takes the surface subsidence monitoring project of a city's foundation pit as an example.First,preprocess the raw monitoring data collected in the survey area,detect the singular values in the survey data,and discard and interpolate the detected singular values to ensure the accuracy of the survey data.Secondly,the DC39-2 monitoring point which has the largest settlement in the survey area is taken as an example to be predicted and analyzed by the support vector machine(SVM)regression model,the grey theory model(GM)and the time series analysis model.The results show that the prediction accuracy of single-model data processing has high and low instability.It can be seen that only using a single model of data processing to predict the results of decision-making,there will be certain risks.Therefore,the paper proposes a method of combinational model prediction and combines the capabilities of the three single models for data mining.The paper also establishes a combined model based on SVM-GM-time series.Through comparing predicting results of models,it shows that the built combinational model can effectively overcome the defects in single model prediction.The model not only improve the accuracy of model prediction,but also make the model prediction more stable and reliable.Finally,using the MATLAB and Eviews software platform,the above monitoring data preprocessing and prediction model establishment are programmed.In order to verify the feasibility of the conclusions in this paper,DC41-2 monitoring point which has the largest subsidence is taken as an example.And the single model and the combined model are used in the same way for processing analysis.It further validates that the combination model used in data processing is superior to any other single model.The corresponding conclusions obtained from the study can provide certain reference values for the foundation pit engineering under similar geological conditions.