基于人工神经网络的电力电子主回路的故障诊断
Fault diagnosis of power electronic circuits based on artificial neural network
本文主要研究了双桥串联可控整流电路的故障分类及诊断方法。对1、2、3只晶闸管故障情况进行了分析，并给出了基于整流电压畸变波形分析的故障分类，总结出了25类故障共294种故障状态。通过对故障电压分段平均值进行快速傅立叶变换，最大限度的消除了控制角对电压波形的影响，同时降低了网络样本的维数。最后通过仿真验证了故障诊断方法的可靠性。
his paper mainly studies the fault classification and diagnosis method of the double-bridge series controlled rectifier circuit. All of the three faults included are analyzed, and the fault classification method according to the rectifier aberrant volta- ge waveforms is put forward, and 25 kinds (294 states) of the fault voltage waveforms are summarized. The sub-area mean value of fault voltage is operated with FFT, and the impact on voltage waveforms by triggering angle is eliminated. Meantime, the sam- pling data dimension is decreased. Lastly, the reliability of the diagnosis algorithm is validated by simulation.
苏建元、王大伟、方健、俞华
变压器、变流器、电抗器电子电路
故障诊断可控整流电路模式识别
fault diagnosiscontrolled rectifier circuitpattern identification
苏建元,王大伟,方健,俞华.基于人工神经网络的电力电子主回路的故障诊断[EB/OL].(2007-11-01)[2025-08-21].http://www.paper.edu.cn/releasepaper/content/200711-27.点此复制
评论