摘要
核函数极限学习机有效地避免了极限学习机(ELM)模型固有的随机性和支持向量机(SVM)模型求解的复杂性,而且具有更快的学习速度和更好的泛化性能。因此,提出了基于核极限学习机的模拟电路故障诊断新方法,描述了电路故障特征的选取过程,建立了以核极限学习机为基础的模拟电路故障诊断模型。实验结果表明,该方法故障诊断准确率大于99%,性能优于支持向量机和极限学习机。
The KELM algorithm with the characteristic of fast learning speed and strong generalization is used to construct soft sensor models; this overcomes the randomization of ELM and the complexity solution process of SVM.So a new method for analog circuit fault diagnosis based on kernel extreme learning machine（KELM） algorithm is proposed in this paper. The method for extracting the fault signatures of the circuit under test is proposed and the analog circuit fault diagnosis model based on KELM is established. The simulation results and their analysis testify preliminarily that the proposed approach for analog circuit fault diagnosis achieves excellent performance,obtaining a fault diagnosis accuracy rate of greater than 99%.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
2015年第2期290-294,共5页
Journal of Northwestern Polytechnical University
基金
航空科学基金(2012ZD53051)资助
关键词
模拟电路
故障诊断
支持向量机
核函数
核极限学习机
analog circuits
fault diagnosis
kernel function
KELM（Kernel extreme learning machine）
