Deformation Analysis Based On Space-time Series Model

Scientific, accurate and timely analysis and predict the deformation of engineeringstructures is important for the safety of construction engineering construction and normaloperation. Choose a reasonable model to fit deformation data and accurately forecast thefuture is the main content of deformation analysis.Time series is the chronological order of the data series, time series analysis is to revealthe statistics from a statistical point of statistical relationships between the timing relationshipwith the internal timing.In recent decades, time series analysis have achieved importantresults. But time series model is only for a single deformation monitoring point sequencemodeling, without considering the correlation between monitoring points.Temporal sequence (Space-Time Series) is the time-sequence spread in space, means acollection of spatially related time series relationship. Temporal data is a combination of timeand spatial data, compared with ordinary spatial data and time series data, temporal sequencedata have temporal and spatial correlation and temporal and spatial heterogeneitycharacteristics. Properties of space and time series data is an important prerequisite fortemporal data modeling, in order to allow time and space relationships and temporal patternsmore clearly reflected, temporal sequence data modeling spatial and temporal dependencemust be considered. How to effectively analyze temporal sequence data to construct temporalintegration of spatial and temporal prediction model has been an important part of space-timedata analysis. In this paper, I researched the nature of space-time data including the temporaland spatial autocorrelation and spatial and temporal stability, described the process oftemporal sequence modeling in detail. A pipeline monitoring point elevation values near adeep pit as for the sample data, first, time series analysis, the autocorrelation function and thepartial correlation function to determine the number of possible models, each model order bycomparing the criteria to determine the optimal model, and short-term forecasts, to pave theway for the space-time series modeling; Secondly followed by the use of space-timeautoregressive moving average model (STARMA) to build the pipeline around foundationelevation values of temporal relationship, for temporal analysis and short-term forecasts. The model considered the location of the predicted value of the time series, but also taking intoaccount the position of the time series of spatially adjacent. Proved the model in foundationsettlement data pipeline and forecasting short-term feasibility by comparing the results withtime series modeling, spatial and temporal analysis of temporal sequences.