H桥逆变器IGBT开路故障诊断方法研究

针对H桥结构的逆变器中功率器件（insulated gate bipolar transistor,IGBT）开路故障的特点,提出了基于小波多分辨率分析、核主成分分析和最小二乘支持向量机的故障诊断方法。此方法选取半个基波周期的电容电压平均值为原始信号,避免了负载变化对故障诊断的影响。首先利用小波多分辨率分析对原始信号进行多尺度分解提取出特征向量,然后利用核主成分分析方法来实现特征降维,最后建立了基于最小二乘支持向量机的故障分类器。在一台660 V低压静止同步补偿器试验样机上进行了试验和分析,结果表明该方法具有良好的准确性和实时性。...

Full description

Saved in:
Bibliographic Details
Published in电机与控制学报 Vol. 18; no. 5; pp. 112 - 118
Main Author 杨晓冬 王崇林 史丽萍
Format Journal Article
LanguageChinese
Published 中国矿业大学信息与电气工程学院,江苏徐州,221116 2014
Subjects
H桥逆变器
故障诊断
最小二乘支持向量机
核主成分分析
绝缘栅双极型晶体管
H-bridge inverter
fault diagnosis
故障诊断
最小二乘支持向量机
核主成分分析
H桥逆变器
绝缘栅双极型晶体管
least squares support vector machine
insulated gate bipolar transistor
kernel principal component analysis
Online AccessGet full text

Cover

More Information
Summary:针对H桥结构的逆变器中功率器件（insulated gate bipolar transistor,IGBT）开路故障的特点,提出了基于小波多分辨率分析、核主成分分析和最小二乘支持向量机的故障诊断方法。此方法选取半个基波周期的电容电压平均值为原始信号,避免了负载变化对故障诊断的影响。首先利用小波多分辨率分析对原始信号进行多尺度分解提取出特征向量,然后利用核主成分分析方法来实现特征降维,最后建立了基于最小二乘支持向量机的故障分类器。在一台660 V低压静止同步补偿器试验样机上进行了试验和分析,结果表明该方法具有良好的准确性和实时性。
Bibliography:A new fault diagnosis method of insulated gate bipolar transistor（ IGBT） open-circuit fault was proposed for the H-bridge inverter based on wavelet multi-resolution analysis（ MRA）,kernel principal component analysis（ KPCA）,and least squares support vector machine（ LS-SVM）. To avoid the influence of load variation on fault diagnosis,the average value of capacitor voltage in a half fundamental period was chosen as the original signal. The features of the original signal were extracted by using the MRA. The feature dimensions were then reduced based on KPCA. Finally,the fault classifier was constructed via the LS-SVM. An experimental prototype of the 660 V STATCOM was built to test the proposed method,and the experimental results demonstrate the high accuracy and efficiency of the method.
23-1408/TM
H-bridge inverter;insulated gate bipolar transistor;kernel principal component analysis;least squares support vector machine;fault diagnosis
YANG Xiao-dong, WANG Chong-lin, SHI Li-ping （School of Information and Electri
ISSN:1007-449X