基于SVM的复杂非线性黑箱系统在线辨识方法

支持向量机(SVM)是一种基于结构风险最小化原理的学习技术,是一种新的具有很好泛化性能的回归方法,本文分析了采用神经网络方法进行非线性系统建模存在的缺点,并将SVM应用于复杂非线性黑箱系统模型的在线辨识当中,理论分析和实验证明,该方法学习速度快,跟踪性能好,泛化能力强,对样本的依赖程度低,比神经网络非线性系统建模具有更好的预测精度.

A Method of on-line Identification for Complex Nonlinear System Based on SVM

Support vector is a learning technique based on the structural risk minimization principle, and it is also a kindof regression method with good generalization ability. This paper analyses the disadvantage of the method used for non-linear dynamical systems identification based on neural networks, and uses support vector machine to model nonlineardynamical systems. Theoretical and simulation analysis indicate that this method has the features of high learning speed,good generalization as well as approximation ability, and little dependence on sample set. The present method has thebetter prediction precision than the approach based on the neural network.