【摘要】 为满足故障诊断的实时性和准确性要求,采用阈值化类内距离的方法,研究了一种快速收敛的半模糊c均值(SFCM)聚类诊断方法.证明了SFCM算法的模糊加权幂指数m在区间(0,1)取值时能实现半模糊聚类,讨论了阈值η对算法的影响并给出了算法步骤.以机载武器控制系统信息通道为诊断对象,采用该方法对通道进行了样本无监督分类验证和故障模式识别诊断试验.结果表明:SFCM算法能对信息通道故障模式进行快速准确的分类识别.更多还原

【Abstract】 A fast converging semi-fuzzy c-mean(SFCM) algorithm based on revised Euclidean distance was proposed to fulfill real-time quality and precision in fault diagnosis.It was proved that semi-fuzzy clustering can be obtained when the index factor m is defined in(0,1).The effect of threshold parameters η on the clustering was investigated,and the steps of the algorithm were given.A fault diagnosis example of the information channels in an airborne weapon control system based on SFCM was given.Experimental test of unsupervised clustering and fault patterns recognition to the information channels were implemented.The results show that the algorithm can recognize the fault patterns of the information channels rapidly and accurately.更多还原