摘要
文章针对目前模拟电路故障诊断中存在的容差和非线性特性所带来的诊断难点,提出了一种基于粗集-主成分分析的模拟电路故障诊断的方法。这种方法利用粗集理论属性约简、数据归一化、主成分分析对输入数据进行预处理,提取主要参数,然后利用神经网络对各种状态下的特征向量进行分类决策,实现模拟电路的故障诊断。对标准电路仿真结果表明:该方法能够实现快速故障检测与定位,具有准确率高的特点。
The tolerance and nonlinearity in analog circuit make its fault diagnosis difficult. In order to solve this problem, an analog circuit fault diagnosis method based on Rough Sets-Neural Network is presented. The method uses Rough Sets attribute reduction ,normalization of data and principle component analysis to process input data and generate major ones. Feature vectors under certain states can be classified using NN, and fault diagnosis is realized. Simulation results on benchmark circuits show that this scheme is feasible and has powerful features, such as diagnosing and locating faults quickly and exactly.
出处
《四川理工学院学报（自然科学版）》
CAS
2009年第4期94-96,100,共4页
Journal of Sichuan University of Science & Engineering(Natural Science Edition)
基金
国家高技术研究发展计划(2008AA11A134)
关键词
模拟电路
故障诊断
粗集理论
神经网络
主成分分析
analog circuit
fault diagnosis
Rough Sets
neural network
principle component analysis
