摘要
模拟电路故障诊断的关键是故障特征提取,使故障特征与故障状态(故障元件组合)间有清晰的对应关系。由于模拟电路软故障的连续性和无限性使故障特征和故障状态间的对应关系变得模糊,增加了软故障识别的困难。针对该问题,本文提出了一种基于多激励响应误差矩阵奇异值分解的奇异角故障特征提取方法。奇异角特征具有在单故障情况下仅与故障部位有关的特性,能有效克服软故障程度的连续性对故障特征的影响;同时奇异角特征能减少测试点数、压缩特征矢量维数和提高故障隔离率。文中重点分析了奇异角特征对单软故障和灾难性多故障的识别性能,并结合实例仿真说明该方法的有效性。
To find out the one-to-one relationship between the faults and the fault features is very important process for analog circuit fault diagnosis. It is more difficult to find this relationship due to both the component＇s tolerance and the continuity of soft-faults, which increases the difficulty in soft fault diagnosis. In this paper, an approach of soft-fault diagnosis based on characteristic extraction with singular-value decomposition （SVD） is presented. Firstly, singular-angle characteristics （SAC） are structured by SVD of the error matrix produced from multi-stimulations. Then, the fault identification abilities in cases of both soft-fault and catastrophic multi-faults are discussed. The primary advantage of SAC is that SAC is only related to the fault location in the circuit structure under the single fault situation, so that it can eliminate the influence of the fault continuity on the fault characteristic. Besides, SAC can reduce the number of the test-points, compress the number of dimensions of the characteristic vector, and increase the fault isolation of different faults. Finally, an example is given to illustrate the validity of the proposed approach.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2006年第9期1100-1103,共4页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(60372001)资助项目
关键词
故障诊断
奇异值分解
奇异角特征
faults diagnosis singular value decomposition singular angle characteristic
