摘要
针对模拟电路的故障诊断,提出一种基于主元分析与极限学习机相结合的方法。该方法利用主成分分析法对提取的特征数据进行降维,再结合极限学习机对电路故障进行分类。主元分析具有数据压缩及特征提取的优点,极限学习机学习速度快、泛化性能好。实验结果表明,采用PCA-ELM结合对故障数据处理,故障诊断分类的准确性可达98.3%以上。
Based on principal component analysis （PCA） and the extreme learning machine （ ELM）, a method of fault diagnosis in analog circuits is proposed. The response feature parameters are preprocessed by PCA to generate the major ones. Feature vectors under certain states can be classified using ELM, and fault diagnosis is realized. The ELM enjoys quick learning speed and good generalization performance and compressing data characteristics of PCA. Simulation results on benchmark circuits show that this scheme is feasible with a fault diagnosis accuracy of over 98. 3%
出处
《电子科技》
2017年第5期72-75,79,共5页
Electronic Science and Technology
关键词
模拟电路
故障诊断
极限学习机
主元分析
fault diagnosis
analog circuits
extreme learning machine
principal component analysis
