摘要
为了提高歼击机故障诊断的准确性与实时性,提出一种基于决策树型组合策略的多重核学习支持向量机诊断方法.决策树型组合策略利用树结构解决多分类问题.而多重核学习支持向量机通过混合核空间,将线性约束下二次规划问题转化为二次约束下二次规划问题.实验结果表明:多重核学习支持向量机的诊断精度明显优于标准支持向量机,且支持向量的数目也较少.决策树型组合策略的引入可以提高歼击机故障诊断的诊断速度.基于决策树型组合策略的多重核学习支持向量机方法能够准确且快速地解决歼击机故障诊断问题.
Based on decision tree combined strategy and multiple kernel learning support vector machines, a new fault diagnosis method is proposed to improve the precision and speed of fighter fault diagnosis. The multi-class classification problem is solved by decision tree combined strategy. And multiple kernel learning support vector machines are able to translate linearly constrained quadratic programming into quadratic ally constrained quadratic programming by mixed kernel space. Simulation results indicate that the diagnosis precision by multiple kernel learning support vector machines is better than the standard support vector machines with less number of support vectors. Moreover, the introduction of decision tree combined strategy guarantees the diagnosis speed improvement. The proposed method can solve the fighter fault diagnosis problem accurately and rapidly.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2007年第A01期1-5,共5页
Journal of Southeast University：Natural Science Edition
基金
国家自然科学基金重点资助项目(60234010)
航空科学基金资助项目(05E52031)
关键词
故障诊断
支持向量机
多重核学习
决策树
fault diagnosis
support vector machines （SVM）
multiple kernel learning
decision tree
