摘要
针对航空发动机的故障样本,提出了一种基于动态聚类的粗糙集规则提取算法。给出了该算法的模型,描述了动态聚类方法和广义欧氏距离,举例说明了这种算法,用神经网络对样本进行训练并验证约简是否正确。结果表明,动态聚类法可以改善分类,使最终的核与约简更精准,去除了干扰信息的影响,在保证诊断精度的同时。提高了故障识别的正确率。
Considering the fault samples of aero-engine,the rule extraction algorithm of rough set based on dynamic clustering method is proposed.Firstly the model of the algorithm is obtained,where the dynamic clustering method and general Euclidean distance are defined,and then a given example illustrates this extraction algorithm,at last a Neural-network is used to train the fault samples,and validates whether the reduction is correct.The result we get shows that the dynamic clustering method can improve the classification,and the ultimate nuclear and reduction are more accurate.With removing the impact of interference information,the proposed method improves the accuracy of fault identification as well as ensuring the precision of diagnosis.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第13期3054-3056,共3页
Computer Engineering and Design
基金
国家自然科学基金委员会与中国民用航空局联合项目(60879017)
天津市自然科学基金项目(08JCYBJC11600)
关键词
动态聚类法
广义重要度
广义欧氏距离
粗糙集
属性约简
dynamic clustering method
general important degree
general Euclidean distance
rough set
attribute reduction
