Application Of Support Vector Machine To Safety Monitoring Data Analysis Of Dam

With the water resources exploitation and utilization, the dam safety problems have become increasingly remarkable. It is important to forecast the monitoring data of dam precisely for supervising and decision-making of the dam operation. The dam safety monitoring model is the main tool of analysis, evaluation of the state of the dam, and in the dam safety monitoring plays an important role, the dam prototype observed data analysis and the establishment of model safety monitoring is the safety monitor work final value manifests.Based on the analysis of the dam monitoring traditional statistic model, on the basis of a new data mining method - Support Vector Machine, this paper built the Support Vector Machine dam monitoring statistic model. The analysis showed that the model apply successfully in dam prototype observed data analysis .The major research results:(1) In the dam safety monitoring modeling analysis, the modeling factors mainly consider water pressure, temperature, aging and other factors, and to build the complex relationship between these factors and effect quantity, Least Squares regression statistic method is a commonly used model method, but this linear statistic model is difficult to reflect this complex relationship. But Support Vector Machine, through the introduction of the kernel function, maps the nonlinear problem of the input space to the high-dimensional feature space and in the high-dimensional space structures linear function.So it is a good kind of nonlinear model. In view of these, this paper built the Support Vector Machine dam safety monitoring statistic model.(2) Support Vector Machine solves a convex optimization problem, computational complexity, the speed slowly, and Least Squares Support Vector Machine uses equality constraints to alternative inequality constraints .So the speed of solution accelerates greatly. So, this paper built the dam safety monitoring regression statistical model based on Least Squares Support Vector Machine .It accelerated the solution speed, and reduced the computational resource, and had promoted strength compared to the traditional Support Vector Machine model(3) In view of multiple correlation among factors and nonlinear characteristics of model in dam safety monitoring. In the paper, the dam safety monitoring model based on the combination of partial least squares regression (PLS) and Least Squares Support Vector Machine (LSSVM) was built. The factors affecting dam seepage was analyzed by means of PLS to extract the most important components, thereby not only the problem of multiple correlation can be solved but also the amount of inputing dimensions of the LSSVM can be reduced. The example showed that PLS-LSSVM was more suitable to the model of large-scale data due to greater training efficiency than LSSVM.