Research On Dam Safety Monitoring Method Based On Attracting Method

Dams play a decisive role in the safe operation of reservoirs.In order to avoid the huge losses caused by dam damage,it is necessary to carry out more perfect safety monitoring of the dam.At the same time,the degree of dam safety should be estimated through the analysis and processing of monitoring data and data.The interaction between dam,reservoir water and dam foundation in operation and the external environment(such as temperature,water pressure and earthquake)will affect the actual behavior of the dam.These factors make the time series of dam monitoring data have high non-linear characteristics.Therefore,in order to analyze and judge the time series of dam monitoring data more clearly and comprehensively.It is necessary to introduce a non-linear theory to model dam monitoring data according to its own law of time series.The main contents of this paper are as follows:(1)Nonlinear judgment method based on attractor eigenvalues and its application in monitoring data analysis.According to the theory of non-linear dynamic system,this paper compares and analyses the test performance of the substitution data method for two kinds of non-linear chaotic time series,Lorenz equation and Henon map,when the time inversion irreversibility and attractor characteristic are used as characteristic statistics respectively under different noise disturbances.It proves that the substitution data method with Hurst exponent as non-linear characteristic shows strong robustness.Finally,it is applied to the time series of displacement monitoring data of three dam stations in practical engineering,and the validity and practicability of the non-linear judgment method based on attractor characteristic quantity in practical engineering are demonstrated.(2)The determination of reasonable length in chaotic identification of dam safety monitoring data series.Aiming at the monitoring data sequence of this practical proj ect,different data lengths are selected when reconstructing phase space,and the rule of chaotic characteristics changing with the length of the sequence is studied.The reasonable data lengths when the attractor characteristic quantity tends to be stable are selected.The results show that when the length of the sequence reaches about 1500,the non-linearity of the sequence is stable,the result is reliable and the calculation time is shortened.(3)A denoising model of dam displacement time series based on attractor characteristic is established.Local projection denoising based on correlation dimension iteration is introduced into time series analysis of dam displacement monitoring.The modeling process and steps are described in detail.The simulation results of denoising Lorenz sequence with noise prove its rationality.Then,the method is applied to the denoising of the horizontal displacement time series of PL1 monitoring points.Compared with the linear wavelet packet denoising method.The results show that the local proj ection denoising method based on correlation dimension iteration can be applied to the analysis of dam displacement monitoring data.The denoising effect is good and has certain practicability.(4)Application of attractor analysis in monitoring data analysis of dam safety monitoring prediction model.In this paper,the attractor analysis method is used to establish the prediction method based on the largest Lyapunov exponent prediction method and Markov chain modified variable dimension fractal-gray prediction model,and the prediction method is compared with the traditional least square prediction model.The results show that the two models have high accuracy and fast calculation,and the variable dimension fractal-gray prediction model modified by Markov chain has higher accuracy,which shows that the combination of multiple models can improve the prediction accuracy and is more suitable for dam safety monitoring data prediction processing.