| Support Vector Machine (SVM) based on the SLT (statistical learning theory) is a new machine learning method, which was developed by Vapnik and his team in 1995; it embodies the theory of structure risk minimization (SRM) and can solve the problem characterized by nonlinear, high dimension, small sample and local minimizing perfectly. SVM has become the hotspot in the field of machine learning because of its excellent learning performance, so it was applied successfully in many engineering fields.This thesis applied SVM to the railway engineering, because it has better learning feature and future apply value which is hoped to solve much problem in thedata mining(DM).The following work was conducted in this thesis:1. Expatiated and compared the common methods in the passenger prediction and investment prediction.Detailedly deduced the training and decision-making process of SVM from linear SVM to Non-linear SVM and sum the training algorithm.2. Summarized the theory of neural network, constructed the passenger volume prediction model based on BP and railway investment prediction model based on RBF by programming on MATLAB language.3. Constructed the regression model on the base of theory of SVM, and then applied this model on the prediction of city railway passenger volume by LIBSVM.Then compared the results between SVM and BP, result indicate that SVM is more precise than NN in the situation of small sample.4. Prediction of the investment is the hotspot of railway project, constructed the support vector regression model by LIBSVM, and applied it on the TBM railway tunnel cost prediction, result show that SVM have a better performance than NN in the situation of high dimension.The results show that SVM method is better than neural network. It is believed that Using SVM theory to solve regression problem is a method with promising prospect. At last, a personal preview of further tasks in the research realms of neural network and support vector machine is presented. |
PDF Full Text Request