基于支持向量机属性约简集成的模拟电路故障诊断  

Analog circuit fault diagnosis based on attribute reduct ensemble of support vector machine

作　　者：马超[1] 陈西宏[1] 徐宇亮[1] 姚懿玲 

机构地区：[1]空军工程大学导弹学院,咸阳713800 [2]空军驻210所军事代表室,西安710065

出　　处：《仪器仪表学报》2011年第3期660-666,共7页Chinese Journal of Scientific Instrument

摘　　要：针对模拟电路故障数据存在大量无关或冗余特征的特点,为进一步提高故障诊断准确率,提出支持向量机属性约简集成的模拟电路故障诊断新方法。首先证明一致决策表属性约简与集合覆盖的等价性,将最优属性约简问题转化成最小集合覆盖问题;然后在结合混沌优化产生初始信息素分布和进行混沌扰动的基础上,设计求解最小集合覆盖问题的混沌蚁群算法;最后给出基于属性约简集成的模拟电路故障诊断模型。用双二次滤波电路对算法进行验证,取得97.8%的故障诊断准确率,与其他方法进行比较,结果显示了本文方法的优越性。Aiming at the problem that there are many irrelated and redundant characteristics in fault data,a new method of analog circuit fault diagnosis based on attribute reduct ensemble of support vector machine（SVM） is presented to enhance the accuracy of fault diagnosis.Firstly,the equivalence of attribute reduction and set-covering is proved,and then the problem of optimal attribute reduction is transformed into the problem of minimum set-covering.Then,the chaos ant colony algorithm is designed to solve the minimum set-covering,which is based on initial pheromone distribution and chaotic disturbance generated both with chaotic optimization method.Finally,an analog circuit fault diagnosis model based on attribute reduct ensemble is proposed.A biquad filter circuit is used to verify the proposed method,and the fault diagnosis accuracy is 97.8%.Compared with other methods,the presented method is superior.

关 键 词：模拟电路 故障诊断 支持向量机集成 属性约简 混沌蚁群优化算法 集合覆盖 

分 类 号：TN707[电子电信—电路与系统]