摘要
支持向量机是一种具有完备统计学习理论基础和出色学习性能的新型机器学习方法,它能够较好地克服过学习和泛化能力低等缺陷。但是在利用支持向量机的分类算法处理实际问题时,该算法的计算速度较慢、处理问题效率较低。文中介绍了一种新的学习算法粗SVM分类方法,就是将粗糙集和支持向量机相结合,利用粗糙集对支持向量机的训练样本进行预处理,通过属性约简方法以减少属性个数,且在属性约筒过程中选出几组合适的属性集组成新的属性集,使模型具有一定的抗信息丢失能力,同时充分利用SVM的良好推广性能,从而缩短样本的训练时间,实现快速故障诊断。对航空发动机故障诊断的实验结果表明了该方法的优越性。
support vector machine is a kind of new machine learning method.This method has good generality capability and better classification accuracy.But when solve real problem using support vector machine,its computation, rate is slow and its efficiency is low.Introduce a kind of method named rough support vector machine （RSVM） that improves the real-time character of prediction sys- tem based on SVM in this paper. RSVM has high classification accuracy with much less attributes, which means less sensors and less cost.And it keeps certain redundant attributes to have high fault diagnosis accuracy in the case of lost information caused by sensor fauh.RSVM increases classification accuracy with good generalization performance.The numerical experiments for aero-engine fault diagnosis show the effectiveness of the proposed algorithm.
出处
《微计算机信息》
北大核心
2008年第31期189-191,共3页
Control & Automation
基金
陕西省自然科学基金资助项目(2007F36)
关键词
粗糙集
支持向量机
航空发动机
故障诊断
rough sets
support vector machine
aero-engine
fault diagnosis
