摘要
提出基于堆叠法的多分类器集成算法,用于降压式变换器电路故障诊断。首先利用蒙特卡洛采样提取出特征参数制作降压式变换器电路数据集;利用不同分类器对训练样本进行五叠交叉验证分类,再根据堆叠法的两层模型将多个分类器融合,提高预测准确度。通过与各算法分类效果对比,本文集成算法提高降压式变换器电路的故障预测准确率达97.8%,验证了算法的可行性及有效性。
This paper proposes a multi-classifier integration algorithm based on the stacking method for fault diagnosis of step-down converter circuits.Firstly,Monte Carlo sampling is used to extract the characteristic parameters to make the circuit dataset of the step-down converter.Different classifiers are used to classify the training samples with five-stack cross-validation,and then multiple classifiers are fused according to the two-layer model of the stacking method so as to improve the prediction accuracy.By comparing the classification effect with other algorithms,the integrated algorithm can effectively improve the fault prediction accuracy of the step-down converter circuits to 97.8%,which verifies the feasibility and effectiveness of the algorithm.
作者
毛伟
余刃
刘明辉
殷少轩
MAO Wei;YU Ren;LIU Minghui;YIN Shaoxuan(College of Nuclear Science and Technology,Naval University of Engineering,Wuhan 430033,China)
出处
《兵器装备工程学报》
CAS
CSCD
北大核心
2023年第S01期364-368,共5页
Journal of Ordnance Equipment Engineering
关键词
降压式变换器
电路故障诊断
分类算法
堆叠法
集成算法
step-down converter
circuit fault diagnosis
classification algorithm
stacking method
integration algorithm
作者简介
毛伟(1980-),男,副教授,E-mail:hanlu_127@163.com;余刃(1968-),男,教授,E-mail:18071068480@163.com;刘明辉(1993-),男,E-mail:byesbuter@163.com;殷少轩(2000-),男,E-mail:1033272793@qq.com。
