摘要
在内燃机神经网络故障诊断系统的基础上 ,引入粗糙集理论 ,对其在内燃机故障诊断特征参数属性优化中的运用进行了探索。利用可辨识矩阵算法对决策表进行属性约简 ,剔除其中不必要的属性 ,揭示了故障诊断条件属性内在的冗余性 ,降低了神经网络构成的复杂性。最后给出了属性约简的结果。
On the basis of ANN fault diagnosis system of internal combustion engine,rough sets theory is introduced,and its application in characterized paramaters attribute optimization is explored too.The discernibility matrix algorithm is used to reduce attributes in decision table.The unnecessary attributes are eliminated.The inner redundancy of fault diagnosis systems condition attributes is revealed.And the complexity of neural networks structure is also decreased.The result of attribute reduction is given finally.
出处
《内燃机学报》
EI
CAS
CSCD
北大核心
2002年第2期153-156,共4页
Transactions of Csice
基金
总参军训部国防预研基金
国家高等学校博士学科点专项基金 (980 61 1 1 7)
