Application Of Multiple Grey Predicting Model For Monitoring Dam Safety

The dam of reservoir contributes greatly to the development of national economy,however,once an accident occurs,it will bring to people great tragedy. So it is of great significance to monitor dam safety and analyze the observation data regularly to evaluate safety status of the dam. In china,a large amount of dams lack complete observation data because of financial shortage or other reasons. When the dam is under construction or on the initial stage of reserving water,long-period observation data is also lacking. At present,some existing models used to analyze observation data of dam monitoring such as multiple regression and neural network are formulated on the basis of a large number of data. In case of short-period or incomplete observation data,they are helpless. Moreover,there are drawbacks like difficulties in stimulation,complexity of calculation and heavy work in all these models because of many other factors. For example,the complexity of the dam structure,difficulties in measuring the physical and mechanical parameters of building materials and subgrade soil,stimulating the geologic structure of subgrade,predicting effects of loads,construction and environment on the dam and so on. In view of the above problems,multiple grey predicting model is proposed to analyze the observation data in the thesis.Based on practical observation data of dam monitoring,examples of multiple grey predicting model are given in this paper. The satisfying results have shown that the method is feasible and effective. To settle the problem of singular matrix existing in the application of multiple grey model,MooreHPenrose inverse of matrix is proposed and an example of this is given. Although the result is satisfying,more practical data are needed to test this model. To improve predicting precision of multiple grey model,number-transformation method is applied. The author thinks that strictly and unitary increasing is not a necessary condition for applying these methods and it agrees well with the results,which have shown that these methods can be applied in small-scale fluctuating data to improve the predicting precision of multiple grey model. Unlike the custom way which suppose the first number as a known one,the boundary condition of multiple grey model is modified and a new predicting formula is proposed. The result has shown that this new predicting formula can improve the precision of multiple grey model.Finally,some suggestions on later study are given.