Application Of Wavelet Analysis And Time Series Analysis Combined Model In Predicting The Deformation Monitoring

With the rapid development of modern technology and the rapid progress of the nationaleconomy, the speed of the process of construction of the modern variety is also greatlyaccelerated, now we have higher requirements engineering building construction scale,accuracy, etc, so in order to ensure safe operation of the project construction work for variousprojects deformation monitoring is particularly important, especially for deformationmonitoring data analysis and processing, is a top priority.At present, for deformation monitoring data processing are mainly concentrated on theanalysis of deformation causes and forecast the future, we in the case of limited observationdata to predict the deflection of the future, there is a big difficulty, now, the general is tochoose the effective mathematical model, according to the temporal characteristics of themonitoring data to carry on the forecast. Now commonly used in deformation monitoring dataprocessing model mainly include: regression analysis model, time series analysis model, graytheory model, artificial neural network model, kalman filtering model, wavelet analysis modeland so on. Of each model has its own advantages and disadvantages, and collection ofdeformation monitoring and deformation factors of many factors, sometimes in thedeformation data contain a variety of factors, such processing data requires the intersection ofmultiple disciplines, the required processing model is varied, the processing precision of thesingle model could not meet requirement. Therefore, we are now commonly used portfoliomodel way to solve this problem, a portfolio model is the use of the advantages of each model,and the organic combination allows it to more effectively handle all kinds of deformationmonitoring data, improve the accuracy of prediction.Based on consulting a large number of documents and materials and all kinds ofengineering examples, proposed the use of wavelet analysis and time series analysis methodsin combination. Time series analysis model is a dynamic model, for all types of deformationmonitoring data has a very good compatibility, but when dealing with non-stationary timeseries data, there is a trend resulting in poor differentiation remove deleted data caused by thelimited accuracy of the reduced forecast problems. Wavelet analysis model wavelet transformis a way to efficiently extract data from the timing error, the wavelet transform through the monitoring data decomposition and reconstruction, can well reflect the deformationmonitoring data trends and characteristics, thereby separating errors. Based on this, thecombination of the two models used to effectively solve the time-series analysis of trends inquestion removed, the combined model in the actual processing of deformation monitoringdata has a good practical value.