Research On Deflection Monitoring Design Of Concrete Gravity Dam Based On Numerical Simulation

Dam monitoring data model is a hotspot and difficulty in dam safety monitoring research.Dam monitoring data models mainly include single monitoring point model and multidimensional multi-monitoring point model.Single monitoring point model has high accuracy and wide application,but it is difficult to reflect the overall characteristics of the dam because of its limited utilization of the overall monitoring data.However,the construction,solution and analysis of multi-dimensional multi-point model are complicated and are still in the process of continuous exploration.The deflection model is a special form of multi-point model,which is seldom studied at present,and it is suggested to analyze the deflection data of concrete dams in the safety monitoring code.Therefore,it is of great theoretical significance and practical value to carry out this research.In addition,this paper is exploratory,with an unknown attitude to understand the dam body deflection,and the numerical model to simulate the real dam,not limited by the real dam monitoring data in quantity and quality.Therefore,based on the numerical simulation dam,this paper proposes the identification method of statistical deflection monitoring model of concrete gravity dam by using multinomial regression,uncertainty analysis,DP optimization algorithm,BP neural network and other multidisciplinary knowledge,and carries out the supporting research of relevant contents.The main research contents are as follows:（1）Recognize the law of deflection:Taking concrete gravity dam as the research object,the general type of deflection statistical monitoring model is constructed based on Taylor polynomial.In order to solve the problem of overfitting of polynomial model,an uncertainty analysis method for the identification of statistical monitoring model of concrete dam deflection is proposed to determine the optimal order of polynomial deflection.On this basis,the uncertainty of statistical monitoring model for deflection of concrete gravity dam is discussed.The numerical simulation of concrete gravity dam and related discussion and analysis show that the optimal order of the statistical monitoring model of dam deflection is mainly affected by the water level and temperature of the upstream,and shows a strong seasonal law.The dam deflection caused by upstream water level is approximately linear distribution pattern,and the dam deflection caused by temperature is approximately parabolic distribution pattern.When the deformation caused by upstream water level and temperature are superimposed in the same direction,the nonlinear degree of the dam deflection will be enhanced.Large monitoring errors will "drown" the nonlinear characteristics of dam deflection,while small monitoring samples will "exaggerate" the nonlinear characteristics of dam deflection.It is found that the increase of slope and intercept of statistical deflection model of concrete gravity dam reflects the tendency of the dam to be unfavorable to the stability of the dam body.（2）Design the layout of deflection monitoring points;Taking the dam deflection curve as the research object,the key monitoring points are identified according to the DP optimization algorithm to obtain the deflection data.In addition,the deflection curves of dam body obtained by uniform layout and linear interpolation are analyzed,and the difference of deflection curves obtained by DP optimization algorithm is discussed.Numerical simulation and related discussion and analysis show that the changes of upstream water level and temperature affect the shape of the deflection curve of the dam body,so the number and location of the selected monitoring points change.In terms of quantity,the maximum number of monitoring points of the dam body deflection is 11,at least 2 points.Generally speaking,the reservation of monitoring sites shows certain regularity.Therefore,in order to obtain better statistical and monitoring data of deflection and be more economical,this paper suggests that sensors should be embedded at 2/8,4/8,5/8 and 7/8 of the dam body（direction from the dam foundation to the dam top）,the dam foundation and the dam top.（3）Deflection space-time application model:According to the horizontal deformation monitoring data of the key monitoring points selected in the optimization layout,the deformation of the single measuring point is predicted by the BP neural network model of the single measuring point time series.Then,the space-time application mode of the deflection statistical model is proposed by combining the deformation of the single measuring point with the deflection statistical model.At the same time,the evolution of the key monitoring points selected in the optimal layout is discussed.The results of numerical simulation and related discussion show that the spatio-temporal prediction model can effectively integrate the monitoring data of each measuring point,and the prediction accuracy is high.During the monitoring period,the key monitoring points show obvious periodic fluctuation with the change of time and season and the joint action of upstream water level and temperature.