Research On Analysis And Forecast Of Deformation Monitoring Data Based On Time-space Model

Deformation prediction is a very important content in deformation monitoring.It is to establish a suitable mathematical model for the deformation data and apply the model to reasonably predict the future deformation value of the deformed body.This is essential for safe construction.ARIMA model modeling first needs to perform differential processing on the deformation data to make it smooth.With the deepening of research,it is found that the difference processing causes the waste of time series trend item information,and the prediction accuracy decreases with the increase of the number of periods;the traditional time series model is mainly used in single-point modeling,that is,only the correlation between the series values in the time domain is considered Without considering the spatial correlation between observation points in the spatial domain.In order to analyze the overall deformation law of the deformable body,it is necessary to explore the modeling method of multiple points in different spatial positions.Aiming at the above problems,this paper proposes an optimized trend term hybrid time series model,which is used to model and forecast deformation data;introduce the theory of spatiotemporal modeling,through analysis and derivation,establish a STARMA model of deformation data,And apply the model to deformation data modeling and forecasting.The completed work is as follows:（1）Visually analyze the settlement data of the auxiliary building of the foundation pit and the settlement data of the new high-rise building,and comprehensively judge that the building is in a uniform settlement state;establish a linear regression model and a nonlinear regression model to fit the deformation trend,and obtain a nonlinear model fit The effect is better,and it is judged that the building is in a safe state according to the characteristic law of the regression model.（2）Establish an ARIMA model based on the settlement data of the newly built high-rise building,and take the settlement data of the auxiliary building of the foundation pit as a verification example.The results show that the ARIMA model has high prediction accuracy,but the long-term prediction accuracy is reduced,and the model is more complex.In view of these shortcomings,an optimized trend item time series model method is proposed to model and forecast the settlement data of new high-rise buildings.The results show that,The optimized model has high prediction accuracy and stable long-term prediction accuracy.The model retains the trend item information,is more intuitive than the traditional model,and can better describe the deformation law.（3）The related theories and methods of spatio-temporal modeling are introduced in detail.A STARMA model is established for the 5 observation points of the newly-built high-rise buildings with adjacent relationship,and the model is used to predict the time and space of the observation sequence at the sixth point.The results show that STARMA The model can complete the spatiotemporal prediction of the deformation data,and the effect is satisfactory.