模拟电路的故障诊断与参数辨识代数方法

发布时间：2018-08-27 10:59

【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。
[Abstract]:In view of the difficulty of fault parameter identification in analog circuits, a fault diagnosis and parameter identification method for analog circuits based on main eigenvalue and Frobenius norm is proposed from the point of view of algebraic algorithm. In this method, the sampling sequence of the output response of the circuit under test is formed into a square matrix; the main eigenvalue and Frobenius norm of the square matrix are solved; the main eigenvalue and the Frobenius norm change trend line corresponding to each device that may fail are obtained from the pre-test simulation. Based on the minimum error criterion, the most likely devices are determined, and fault location and parameter identification are realized. The experimental results show that this method can integrate fault location and parameter identification. Compared with some classical algorithms, this method has the advantages of high computational efficiency, high accuracy of fault location and high precision of parameter identification.
【作者单位】：宜宾学院物理与电子工程学院;电子科技大学自动化工程学院;
【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)
【分类号】：TN710

【相似文献】