支持向量机回归在矿区GPS高程转换中的应用

In order to refine quasi-geoid in mining area, a Support Vector Regression(SVR) model based on statistical lerning theory and Support Vector Machines(SVM) principle was employed for GPS height transformation. Besides, SVR, polynomial and GA-BP neural network were proposed to transform GPS height. The compared and analyzed test results show that the precision of SVR is superior to that of polynomial and GA-BP. Moreover, SVR can effectively overcome many shortcomings that neural network exists, such as topology selecting difficult, over-learning, unable to avoid regional maximum etc.