摘要
该文引入压缩感知理论,有效降低了IGBT状态监测系统的采样频率。对功率半导体器件而言,各种电气参数的准确测量是保障其状态监测系统预测精度的重要基础,其中关断峰值电压作为判别IGBT老化状态的重要参数之一,其获取方式严重依赖于高速率采样单元。为了大幅度降低信号采集过程所需的数据量,引入压缩感知理论对信号进行压缩采样和重构。由于关断电压波形在时域下不具备显著的稀疏性,为此引入傅里叶基和离散小波基对信号进行稀疏表示,并讨论其在不同变换基下的稀疏性。此外为了便于压缩感知理论实际工程的实现,对比分析不同观测矩阵下信号的重构精度。最后通过仿真和数据分析,证明了压缩感知理论应用于基于关断峰值电压的状态监测系统的可行性和准确性。
In this paper,the compressed sensing theory was introduced to effectively reduce the sampling frequency of IGBT condition monitoring system.For power semiconductor devices,accurate measurement of various electrical parameters is an important basis to ensure the prediction accuracy of the condition monitoring system.Turn-off peak voltage is one of the important parameters to judge the aging state of IGBT,and its acquisition method relies heavily on the high-speed sampling unit.In order to greatly reduce the amount of data needed in the process of signal acquisition,compressed sensing theory was introduced to compress sampling and reconstruction of signals.Since the turn off voltage waveform has no significant sparsity in time domain,Fourier basis and discrete wavelet basis were introduced to represent the signal sparsely,and their sparsity under different transform bases was discussed.In addition,in order to facilitate the practical engineering implementation of compressed sensing theory,the reconstruction accuracy of signals under different observation matrices was compared and analyzed.Finally,through simulation experiments and data analysis,the feasibility and accuracy of applying compressed sensing theory to the condition monitoring system based on turn off peak voltage were verified.
作者
李庭
陈杰
于泓
邱瑞昌
刘志刚
Li Ting;Chen Jie;Yu Hong;Qiu Ruichang;Liu Zhigang(School of Electrical Engineering Beijing Jiaotong University,Beijing 100044 China;Beijing Engineering Research Center for Electrical Rail Transit,Beijing 100044 China)
出处
《电工技术学报》
EI
CSCD
北大核心
2022年第S01期163-171,共9页
Transactions of China Electrotechnical Society
基金
中央高校基本科研业务费专项资金资助项目(2018YJS163)。
关键词
绝缘栅双极型晶体管
状态监测
压缩感知
稀疏基
Insulated gate bipolar transistor
condition monitoring
compressed sensing
sparse basis
