Deformation Monitoring Model Of Xin'anjiang Gravity Dam Based On Support Vector Machine

When the dam had been built, it will slowly aging with time, that will bring all kinds of potential safety risks and influence the dam’s "health" conditions. Therefore, the dam safety monitoring gradually has been paid attention to, and it is very important to in-depth study the theory and method of the dam and dam foundation.In recent years, a variety of new theories and methods of genetic algorithms, gray theory, catastrophe theory, artificial neural networks have been gradually applied to dam safety monitoring data analysis, the research and application of these theories, which greatly enriched and developed the numerical model method of dam safety monitoring and analysis, also prompted further development of China’s monitoring data analysis. Support vector machine (SVM) is a kind of based on statistical learning theory developed on the general effective new machine learning method, which has good generalization ability, and can effectively overcome the curse of dimensionality, the learning and training process can avoid local optimal solution. This paper mainly takes the monitoring data of the Xin’anjiang’ gravity dam as an example, the principle of SVM is applied to dam deformation monitoring and analysis, and the SVM model of dam deformation monitoring is established.In this paper, the main research contents are as follows:1.Based on the analysis of the basic monitoring data of the Xin’anjiang’gravity dam. to determine the input factors of Xin’anjiang’gravity dam’s deformation monitoring SVM model.2.Based on the libsvm toolbox, the SVM model parameters c and g are optimized by using the K-fold Cross Validation method, and we selected RBF and polynomial two different kernel functions to establish the deformation monitoring SVM model of 16# dam section. After comparison and analysis of the results, it shows that based on the RBF kernel function of the SVM model has a better prediction effect.3. At the same time, the BP neural network model of 16# dam section of Xinanjiang’ gravity dam is established. By comparison and analysis of SVM model and BP neural network model, SVM deformation monitoring model of the complex correlation coefficient, residual square sum, the absolute maximum value, the minimum value, the mean absolute error, the average relative error were 0.894,26.141,2.994,0.003,80.184%,0.989%, BP neural network model of the multiple correlation coefficients, residual sum of squares and the absolute value of residual maximum, minimum, mean absolute error were 0.891,1.315,36.306,3.688,0.064,94.471%,3.161%. The results show that the SVM model is superior to the BP neural network model in the aspects of modeling and forecasting performance.4.Taking the 18 dam section of Xin’anjiang’ gravity dam as an example, to establish SVM model and BP neural network model. After comparison and analysis of the two models, the prediction accuracy of the SVM model is better than that of BP neural network model, To further verify the validity and superiority of support vector machine method.