Application Of Time Series Analysis To The Building's Deformation Monitor

The progress of science and technology, and the development of economics promote the urbanization of China. For the raised material civilization and architectural design and construction skill reaching perfection day by day, as well as the increasingly distinct contradiction between the decrement of land resource and the increment of population with each passing day, there are more and more high-rises in our cities. In order to ensure the service life and security of constructional works, and to demand diagnose, forecast, law and research, the necessity and essentiality to monitor the constructional work and its groundwork systematically in a long time is very clearer. The final purpose of deformation monitor is forecast, namely, to make out the analysis and predict about the future behavior of the constructional work in time by holding the variety rule using long-term accumulated surveying datum. At present, experts of many countries have made deeply researches in this aspect, formed a set of mature theoretical system, and many models to forecast deformation been developed, such as regression model, smoothing model, data series model, and so on. Time series analysis method adopts the parameter model. First it builds a time series model using dynamic data, and then obtains the statistical property of the dynamic data by such parameter model. In this process something most important is this method combines the parameter model with system analysis directly. Though time series analysis method is used widely in every walk of life, it applied less in the field of forecasting constructional work deformation, especially about steel structure buildings. The author does some exploratory work in this aspect. In the thesis the theory of time series analysis, several models about time series analysis and the dynamic feature of time series analysis are expatiated in detail; how to use autocorrelation function and partial correlation function to judge the model is analyzed; a kind of better rule to determined rank is determined by comparing several kinds of determined rank; thereby the reverse function can be used to forecast. In the thesis the author concretely introduces the process to build the time series analysis model by the deformation datum about ChangChun Branch of China Everbright Bank business building(steel structure high-rise), calculates the results of every observation points in the first step forecast, the second forecast and the third forecast, and analyzes the results in detail. The time series analysis method as a kind of method to perform dynamic deformation analysis is feasible to deal with the monitor datum. The observation series can be fitted and forecasted with the higher precision, and the calculation expression is so simple, also the analysis result is reliable. Time series analysis method reveals the statistical relation between the internal statistical relation of each time series and each time series. It is in need of many observation data, furthermore, the determined rank is a relatively complicated process in itself. In addition, the error is inevitable in surveying data, also maybe any abnormal data lie in the observation data, all of this becomes more difficult to determine rank. Therefore the author presents a improved method ——least squares collocation. In the thesis the built method and it's apply of variance-covariance function in least square reckon and estimation method is explained. The prediction result is calculated using the same sample datum, and it is compared with the result of time series analysis. From the compared result the least squares collocation. is verified useful in sedimentation forecast. Using least square reckon and estimation method the prediction precision is less than using time series analysis, but it only need several period datum. Furthermore, the calculated course is very briefness and the regularity is better, and all of this is easy to carry out by computer. So at the present speaking, least squares collocation. is a better method to deal with sedimentation datum and to forecast. The research has not ever been reported in the area. So there is extensive space for studying and developing。...