Application Of Time Series Analysis Combined With EM Algorithm In Settlement Data Processing

During the process of measurement,sometimes,it will resulting in the data of observation missing or make errors for the natural environment or human factors,thus,the whole set of data into incomplete data.If the missing data is indispensable,the time series analysis of the incomplete data modeling analysis of the parameters obtained by the deviation will occur,making the forecast value of a larger error,misleading engineering safety judgments.This paper aims to using the time series to analyze the incomplete data in the monitoring data,and introducing Expected Maximum Algorithm(EM Algorithm),put forward combined EM algorithm indirect adjustment model,combined EM algorithm with time-series analysis.Using the time series analysis of EM algorithm to model the incomplete settlement data encountered during the settlement process,For full data use is time series analysis for modeling analysis,BP neural network modeling analysis the incomplete data,compare the modeling results.The above theory can be applied to the Zhengzhou Metro Line 5 Zheng Bian Road pit settlement data processing,the time series analysis of EM algorithm can solve the problem that time series is not accurate for incomplete data modeling The absolute error is within 0.25mm.EM algorithm for time series analysis of the modeling accuracy is lower than the time series on the complete data modeling accuracy of about 0.05mm.BP neural network mode(?)g accuracy is lower than the EM algorithm time series analysis of the modeling accuracy of about 0.20mm.Note that the time series analysis combined with the EM algorithm can handle incomplete settlement monitoring data with high accuracy and reliable prediction.