Research On The Model Of Earth-Rock Dam Based On Safety Monitoring Data

Statistical models play a major role in analysis of earth-rodk dam safety monitoring.As a result of the shortage of the existing statistical models,the researches based on the safety monitoring data of earth-rock dam are as follows:(1)In order to avoid the phenomena which may occur in the process of data analysis,principal component analysis is used to extract principal component under the condition of minimum information loss.Then the model is used to predict and its predicted effect will be tested.(2)In order to solve the problem of deficiency for monitoring data,autoregressive integrated moving average model is established for nonstationary time series to predict and its predicted effect will be tested.(3)In order to avoid suprious regression which may occur in regression model,cointegration theorem is used to test whether a set of monitoring variables are stationary time series.Vector error correction model which is based on the cointegrated variables is established to predict and its predicted effect will be tested.The conclusions are as follows:(1)2 principal components which contains 99.403 percent of the original information are extracted according to the principal component analysis.Contrasted with the stepwise regression model,principal component analysis overcomes multicollinearity.As a result of this,the R square for principal component analysis is higher and the regression coefficient is more reasonable.(2)In terms of prediction accuracy among the 3 models,the autoregressive integrated moving average model is higher than the principal component regression model and lower than the vector error correction model.(3)Cointegration relationship is confirmed between independent variables and dependent variables.As a result of this,vector error correction model is established based on the cointegration relationship among them.The vertical displacement can be initially judged steadily during the monitoring period.