Modeling Seasonally Monitoried Series

In the analysis of deformation, people have learned that the purpose for deformation is not the deformation measurement, but scientific analysis and forecast. However, the deformation because of seasonal factors can not be neglected. The seasonal components may come from the cycle deformation of the object which enhance the difficulties in the analysis and forecast of the deformation monitoring. Based on the characteristics of seasonal monitoring series, the paper studies the methods of removing and separating the seasonal components from the monitoring series, and building the models, which are relating to single-point and multi-point series to forecast monitoring series with seasonality. Some raw materials such as GPS network materials in South California, gound water materials in Eastern China region and a data series of continuous measurement of a joint meter installed in a concrete structure are integrated into above-mentioned models and methods to compare and validate the efficiency of them in detail. The paper provides references and theoretical bases to analysis and solve these seasonal monitoring series.The main work of the paper concentrates on the following four parts:(1) Introducing the economics terms: seasonal index, to the survey domain, proposing the hybrid grey model: Seasonal index-GM(l,l) to forecast the monitoring series data sets with seasonality;(2) Introducing X-11 used in the business forcasting in American Bureau of Statistics to the survey domain, developing the combined X-11 and ARIMA model, introducing "asymmetry moving average" to improving X-11-ARIMAmodel;(3) Analysising the standard normal distribution (SND) based deseasonalization method, proposing the combined SND and GM(1,1)model;(4) Building the multi-point SSCGM(l,m) model which extends the single-point GM(1,1) model. It illustrates that the deformation of points in monitoring network is correlative, any point is not isolated, they affect each other. The relationship among the monitoring points is considered in the multi-point SSCGM(l,m) model, so it is a superior model from the view of modeling mechanism.