项目作者： fish-kong
项目描述 ：
The fault diagnosis method of 24 pulse rectifier based on machine learning is studied. Firstly, the simulation model of pulse rectifier is built to collect the output current waveform under various faults. Secondly, the fault eigenvectors embedded in the current waveform are extracted by fast Fourier transform and wavelet energy spectrum transform. Finally, BP network and ELM network are used. Fault classification model training. The results show that the use of ELM combined with fast Fourier transform can achieve 100% diagnostic accuracy.研究了基于机器学习的24脉波整流器故障诊断方法。首先对脉波整流器搭建仿真模型采集各种故障下的输出电流波形，其次，利用快速傅里叶变换与小波能量谱变换提取嵌入在电流波形中的故障特征向量；最后利用BP网络与ELM网络进行故障分类模型训练。结果表明，利用ELM结合快速傅里叶变换能达到100%的诊断精度
高级语言：
项目地址: git://github.com/fish-kong/fault-diagnosis-of-24-pulse-rectifier.git
