Study On Settlement Prediction Of Earth Rock Dam Based On Gaussian Process Regression Model

In today’s rapid development of water conservancy,safety monitoring is essential from the beginning of construction to completion,as well as during the operation of hydraulic structures.Only after mastering the deformation of the building,can we understand its real operation state and ensure the safety of hydraulic structures.Therefore,it is particularly important to monitor the deformation of large hydraulic structures and analyze the data.Through the continuous research and unremitting efforts of scholars at home and.abroad for many years,the theoretical model and calculation method of dam safety monitoring direction have made considerable progress and development,which provides reliable theoretical support for the safe operation of the dam.However,objectively speaking,there are still deficiencies in some aspects,which need to be improved.As a new machine learning method,Gaussian process gives the explanation of prediction model,and provides the selection of prediction model and the structure of learning framework.In recent years,the theory of Gaussian process has been developed and widely used in many fields.In this paper,the settlement displacement of earth rock dam is taken as the research object and the Gaussian process regression model is introduced.The main research contents are as follows:(1)This paper introduces the calculation principle and operation flow of the two traditional modeling methods,which are multivariate regression analysis and artificial neural network method.Based on the observation data of water level,temperature and time from 2005 to 2012,the settlement value of a survey point in the project from 2005 to 2012 is fitted,and the settlement deformation in 2013-2014 is predicted.The results of the prediction are compared with the actual settlement,and the results show that the accuracy of the multiple linear regression model and the artificial neural network model is not ideal.(2)The paper introduces the Gauss process regression to the problem of the traditional model which is not accurate enough,and introduces the calculation principle of the Gaussian process regression and the process of establishing the Gaussian process regression model in detail.According to the observation data,the settlement deformation of the actual project in 2013-2014 is predicted,and the prediction results are compared with the results of the multivariate linear regression model and the artificial neural network model,and the accuracy of the prediction of the Gaussian process regression model is verified.(3)In order to further improve the accuracy of the model and reduce the adverse effects of system noise and observation noise on the observation data in complex environment,the wavelet theory is introduced and the calculation principle and operation flow of wavelet denoising are introduced in detail.The model is improved by using wavelet analysis to denoise the observed data of the actual project,and the Gaussian process regression model is established to fit and forecast,so that the model accuracy can be improved.It is proved that the Gaussian process regression model based on wavelet denoising has the feasibility and practicability in the practical engineering application.Gaussian process regression model is not only superior to traditional model in accuracy,but also can give a high interpretable confidence while fitting and forecasting more accurately,which is conducive to judge the decision risk relying on prediction results and provide scientific basis for dam safety operation in the project.