基于最大异类距离和正则极端学习机的模拟电路在线故障诊断

发布时间：2018-09-06 10:06

【摘要】：针对模拟电路在线故障诊断问题,提出了一种将最大异类距离和正则极端学习机结合起来的新方法。首先,利用最大异类距离对初始特征样本进行特征提取,压缩样本规模,获取维数更低、可分性更好的特征样本集;然后,将提取的特征样本送入正则极端学习机进行训练,再利用训练好的正则极端学习机对待测电路实时状态进行诊断,并不断根据诊断结果和新旧样本之间的相似度更新训练样本集;最后,将所提方法用于模拟电路在线诊断中。结果表明:所提方法能够有效实现模拟电路单、双故障的在线诊断,比ELM效果好。
[Abstract]:In order to solve the problem of on-line fault diagnosis of analog circuits, a new method which combines the maximum outlier distance with the regular extreme learning machine is proposed. Firstly, the initial feature sample is extracted by using the maximum heterogeneity distance, and the sample size is compressed to obtain the feature sample set with lower dimension and better separability. Then, the extracted feature sample is sent to the regular extreme learning machine for training. Then the trained regular extreme learning machine is used to diagnose the real time state of the test circuit, and the training sample set is constantly updated according to the diagnosis results and the similarity between the new and the old samples. Finally, the proposed method is applied to the on-line diagnosis of analog circuits. The results show that the proposed method can effectively realize the on-line diagnosis of single and double faults of analog circuits, and the effect is better than that of ELM.
【作者单位】：绥化学院电气工程学院;华南农业大学电子工程学院;
【基金】：广东省高等学校优秀青年教师培养计划资助项目(Yq2013028) 现代农业产业技术体系建设专项资金(CARS-27) 黑龙江省高等学校教改工程项目(JG2014011118)
【分类号】：TN710;TP181

【参考文献】