Gaussian Process Regression Model For Dam Safety Monitoring

According to some data obtained from the dam health monitoring system,several calculation methods were used in the dam safety monitoring procedure to establish a dam monitoring model that can effectively reflect the relationship between the action effect set and the load set.Based on these models,the operational status of the dam can be systematically simulated and evaluated,which can comprehensively assess the real condition of the dam and this is the most common measure to ensure that the dam can be operated in a relatively healthy and efficient status.In the light of the considerable researches conducted by the scholars in this field,many theoretical and calculated models has been proposed,which provided reliable theoretical support for safe operation of dams worldwide.However,there are still some deficiencies need to be improved.The main subject of this paper is concrete dams.Based on the normal monitoring statistical model,Gaussian process regression model was introduced to eliminate the exsiting deficient.Besides,combining with the actual engineering examples,the correctness of Gaussian model has been verified and several other common calculation methods were also introduced to contrast with the above-mentioned Gaussian model.According to the comprasion results,it can be found that the Gaussian process regression model has great advantages in terms of accuracy and ease of selection and optimization of model parameters.The content of this paper not only has a certain guiding effect on the actual project,but also has important significance for the further research of this field in dam safety monitoring in China.The main research work of this article is listed as follows:(1)The statistical models used in the field of dam safety monitoring and the corresponding results were summarized in this study.Comparative analysis was conducted by comparing prediction results obtained from a number of commonly used monitoring models,i.e.,multivariate linear regression model,stepwise regression model,RBF neural network model,with the measured data of a practical project.The insufficiency of the traditional monitoring model in terms of performance was pointed out from the simulated data according to various algorithms.(2)The Gaussian process regression model for dam safety monitoring is proposed based on the Gaussian process regression algorithm,and a horizontal displacement prediction model for a practical project is established by the model.Compared with the results of multivariate linear regression model,stepwise regression model and RBF neural network model,the Gaussian process regression model has advantages in higher accuracy of prediction and the convenience of the selection and optimization of model parameters,which largely make up for the deficient in the traditional monitoring model.