论文部分内容阅读
通过对径向基核函数进行分析后发现:根据样本各个特征的识别能力赋予其不同大小的核参数,可以提高支持向量机的推广能力。此结论基础上,提出了一种基于遗传算法的多核参数径向基支持向量机算法,通过遗传算法最小化验证误差,实现了根据各个特征的识别能力赋予其不同大小的核参数。将该算法用于轴承故障诊断,实验结果表明,与传统支持向量机相比,多核参数径向基支持向量机具有更好的推广能力,同时,核参数的大小反映了对应特征识别能力的大小。
By analyzing the radial basis function, it is found that the kernel parameter of different size can be improved according to the recognition ability of each feature of the sample, which can improve the promotion ability of support vector machine. Based on this conclusion, a genetic algorithm based radial basis function support vector machine (SVM) algorithm with multi-core parameters is proposed. The genetic algorithm is used to minimize the verification error, and the nuclear parameters of different sizes are given according to the recognition ability of each feature. The experimental results show that compared with the traditional support vector machines, the multi-core radial basis function support vector machine has a better ability to popularize. At the same time, the size of the kernel parameter reflects the size of the corresponding feature recognition ability .