支持向量机(SVM)的参数取值决定了其学习性能和泛化能力。对此,将SVM参数的选取看作参数的组合优化,建立组合优化的目标函数。采用蚁群算法(ACA)来搜索最优目标函数值。ACA是一种优化搜索方法,具有较强的鲁棒性、优良的分布式计算机制。仿真表明,ACA是选取SVM参数的有效方法,应用到函数逼近时有优良的性能。 Support vector machine (SVM) parameter values ​​determine the learning performance and generalization ability. In this regard, the selection of SVM parameters as a combination of optimization, the establishment of combinatorial optimization objective function. Ant colony algorithm (ACA) is used to search the optimal objective function value. ACA is an optimized search method, with strong robustness and excellent distributed computing mechanism. Simulation shows that ACA is an effective method to select SVM parameters and has good performance when applied to function approximation.