支持向量机 (SVM)是一种新的通用学习机器 ,它从结构风险最小化的角度 ,分析了学习过程的一致性、收敛速度等。SVM能以任意精度逼近一类函数 ,而与输入的维数无关 ,克服了传统神经网络用于系统辨识的维数灾问题及结构难以确定等缺点。基于这一特性研究了对非线性动态系统的辨识问题 ,仿真结果表明SVM用于系统辨识有良好的辨识效果 ,并指出了今后研究的方向 Support Vector Machine (SVM) is a new general learning machine, which analyzes the consistency of learning process and convergence speed from the perspective of minimizing structural risk. SVM can approximate a class of functions with arbitrary precision, but has nothing to do with the input dimension. It overcomes the disadvantages of traditional neural networks, such as dimensionality disaster and structural uncertainty. Based on this characteristic, the identification of nonlinear dynamic systems is studied. The simulation results show that SVM has a good identification effect for system identification and indicates the direction of future research