SVM是一种新型的机器学习方法,其分类性能的优劣主要受核函数及核参数的影响,国内外学者针对SVM核参数的选择已提出许多算法.本文首先分析了RBF核参数对SVM分类性能的影响,然后又对比分析了目前存在的几种基于RBF核的SVM核参数选择方法.通过实验,发现使用遗传算法选择核参数的SVM有比较快的搜索速度. SVM is a new type of machine learning method, whose merits of classification are mainly affected by kernel function and kernel parameter, many algorithms have been proposed by domestic and foreign scholars for the selection of SVM kernel parameters.Firstly, this paper analyzes the effect of RBF kernel parameter on SVM classification And then compare and analyze several existing kernel parameter selection methods of SVM based on RBF kernel.Experiments show that SVM with kernel parameters selected by genetic algorithm has a faster searching speed.