摘要
针对变流器电路中电力电子器件的故障特性,利用小波分析对采集的数据进行去噪,运用随机森林算法对变流器电路故障类别进行诊断,并与单一随机森林故障诊断方法进行对比。仿真与实验结果分析表明,提出的方法在变流器故障诊断中具有较强的抗噪能力和较高的正确诊断率,尤其在干扰噪声较大时效果明显,在实际工程解决变流器电路故障诊断问题上具有良好的实用价值。
Aiming at the fault characteristics of power electronic devices in converter, using wavelet threshold method to reduce the noises of the collected data, diagnosing the fault classification of converter circuit with random forests, and making a comparison with single random forests diagnosis method. The testing results show that the method owns good ability to resist noises and high accuracy, the effect is obvious when the interference noise is large. The new method has high application value to solve the problem of power electronic circuit fault diagnosis.
出处
《电气技术》
2016年第6期36-40,46,共6页
Electrical Engineering
关键词
变流器电路
小波分析
随机森林
故障诊断
converter circuit
wavelet threshold
random forests
fault diagnosis
