Selection moisture forecasting model kernel function and parameter based on support vector machine

On the basis of introducing the basic principles of Support Vector Regression Machine (SVR) and the validation of underground water forecasting with IP method, in order to improve the prediction accuracy of prediction models and model calculation speed, we select sounding results near the well and pumping test data from Xi Mazhuang water source as the research object, using cross-validation, ladder search and grid search method to determine the support vector regression model parameters. We compare the polynomial, RBF and Sigmoid kernel function, and select the best kernel functions and parameters for the prediction underground water content model based on support vector regression machines using electrical sounding method.