The Research Of System Modeling On Support Vector Machines

The math model is used to image the essential relation of every variable in the process, maybe the model is algebra equation, or differential equation, or geometry curve. The aim of system modeling is confirming the relation of input and output in the math model. Now the hot point of system modeling is nonlinear-system modeling based input and output. There are new methods to settle the question, such as neural networks and wavelet networks. But they are not satisfied us. Support Vector Machines is a new kind of machine learning algorithm based on the statistical learning theory. Because of its standout learning capability, the algorithm has been one of the hot points in international machine learning research. In this paper, Support Vector Machines is presented by choosing different kernel function. In this way, ARMA model, BM model and NARMA model is identified. The favorable identification capability of Support Vector Machines is proved through simulation. It can solve the over-learning question of neural networks.