模拟电路的故障诊断与参数辨识代数方法  

Algebraic method for fault diagnosis and parametric identification of analog circuits

作　　者：周启忠[1,2] 谢永乐[2] 徐娟[1] 

机构地区：[1]宜宾学院物理与电子工程学院,四川宜宾644000 [2]电子科技大学自动化工程学院,四川成都611731

出　　处：《哈尔滨工程大学学报》2017年第4期595-601,共7页Journal of Harbin Engineering University

基　　金：国家重点基础研究发展计划(2014CB744206);国家自然科学基金项目(61371049);校博士基金项目(2016QD06)

摘　　要：针对模拟电路难以实现故障参数辨识的问题,从代数算法角度,提出一种基于主本征值和Frobenius范数的模拟电路故障诊断与参数辨识法。该方法把被测电路输出响应的采样序列组成一个方阵;求解该方阵的主本征值和Frobenius范数;根据测前仿真获得的每个可能发生故障的器件相对应的主本征值和Frobenius范数变化趋势线,以最小误差为判据,确定最可能发生故障的器件,实现故障定位和参数辨识。实测结果表明:该方法可将故障定位和参数辨识一体化处理,与已有的几种经典算法相比,具有计算效率高、故障定位准确率高、参数辨识精度高的优点。To solve problems regarding fault parameter identification in analog circuits,a method for fault diagnosis and parametric identification of analog circuits based on dominant eigenvalues and Frobenius norm was proposed in the point of view of algebraic algorthm.A matrix was built using the sampled time series from the output of the circuit under test(CUT).The dominant eigenvalues and Frobenius norm corresponding to this matrix were obtained.According to the trend lines of the dominant eigenvalues and Frobenius norm of each component,which were obtained based on the simulation before test(SBT),the component most likely to fail was identified.The faults were located and parameters were identified in terms of the least error criteria.Experimental results indicate that the proposed method can simultaneously locate the faults and identify the parameters.Compared with several known methods,the proposed method performs well in fault location and parameter identification and has lower computation costs.

关 键 词：模拟电路 参数辨识 故障诊断 本征值 FROBENIUS范数 故障定位 

分 类 号：TP206[自动化与计算机技术—检测技术与自动化装置]