Monitoring Model of Peak Recognition by Using Dam Monitoring Data of Automatic Monitoring System
The peak value and vale value are more significant than other values of dam monito-ring data. Because the difference of sample size between peak or vale value and ordinary value of monitoring data has not been taken into account, it brings on the lower accuracy of fitting and forecasting of the peak value and the vale value than those of ordinary monitoring data in conventional dam safety monitoring model which are established with the method of least square regression. In order to improve the accuracy of fitting and forecasting of peak value and vale value, peak recognition theory is adopted in this paper. Peak recognition theory needs more data and higher precision. Because of higher frequency, automatic monitoring for dam safety provides us adequate data and chance. The new models are established in which larger weights are given to the peak value and the vale value of monitoring data according to the corresponding sampling frequency proportion and range. On the basis of the theory men-tioned above, conventional stepwise regression model and BP artificial neural network model are improved with peak recognition theory, and monitoring data of typical dams, such as Baishi RCC gravity dam, Bikou earth-rock dam with clay core and Lishimen double-curvature arch dam, are used to validate the method mentioned above. The results show that the accu-racy of fitting and forecasting of measured peak value and vale value of the model has been improved remarkably.
Peak value error safety monitoring weight values fitting forecasting auto-matic monitoring system
Fang Weihua Xu Guolong
College of Water Conservancy & Hydropower, Hohai University, Nanjing 210098,China Nanjing Automation Nanjing Automation Institute of Water Conservancy and Hydrology, Nanjing 210012,China
国际会议
International Conference on Dam Safety Management(2008水库大坝安全管理国际研讨会)
南京
英文
690-693
2008-10-22（万方平台首次上网日期，不代表论文的发表时间）