| 周志官，郭 韵，李渴望.改进核函数的支持向量机智能诊断方法研究[J].轻工机械,2016,34(5): |
| 改进核函数的支持向量机智能诊断方法研究 |
| Intelligent Diagnosis Method of Improved Kernel Function SVM |
| |
| DOI：10.3969/j.issn.1005-2895.2016.05.005 |
| 中文关键词: 智能诊断 支持向量机( SVM) 复合核函数 模拟退火算法 |
| 英文关键词:intelligent diagnosis support vector machine ( SVM ) composite kernel simulated annealing algorithm |
| 基金项目:上海工程技术大学科研创项目（El-0903 -15 -01005 -15 KY0105）。 |
|
| 摘要点击次数: 1333 |
| 全文下载次数: 235 |
| 中文摘要: |
| 针对目前支持向量机( SVM)智能诊断方法核函数选择困难以及参数选择具有随意性的问题，提出了基于模拟退
火算法改进核函数的SVM智能诊断方法，重新设计了支持向量机的核函数以及参数。多项式核函数是局部核函数具有
较强的拟合能力，而径向基核函数是全局核函数具有较强的外推能力，根据Mercer理论，建立一种由多项式核函数与径
向基核函数组合而成的复合核函数，并利用模拟退火算法全局寻优的优点，对支持向量机的参数做最优选择；改进后的
SVM运用于轴承故障诊断。研究结果表明：相对于传统SVM法，该方法具有较好的学习效率及较高的诊断准确率；该方
法运用于轴承故障诊断领域极大地提高了故障诊断的准确率以及诊断效率。该研究为基于模拟退火算法改进核函数的
SVM智能诊断方法应用于机械设备故障诊断提供了相应的理论和实践依据。 |
| 英文摘要: |
| Aiming at intelligent diagnosis method of support vector machine kernel function parameter selection and
random selection difficult, intelligent diagnosis method proposed SVM kernel function based on improved simulated
annealing algorithm, and redesigned SVM kernel function and parameters. Polynomial kernel function is localized
compound kernel function which has a strong ability to fit, and the radial basis function is a global kernel function which
has a strong extrapolation. According to Mercer theory, the establishment of a combination of polynomial kernel and
radial basis function from the kernel, and simulated annealing algorithm for global optimization of the advantages of SVM
parameters were the best choice. Improved SVM bearings were used in fault diagnosis. The results show that compared
with the traditional SVM, the method has better learning efficiency and high diagnostic accuracy. The method used in
fault diagnosis bearing greatly improves the accuracy and efficiency of diagnosis fault diagnosis, and it provide theoretical
and practical basis for mechanical malfunction diagnosis. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
