基于小波包能量法及神经网络的电力电子装置故障诊断

Fault Diagnosis of Power Electronic Device Based on Wavelet Packet Energy Method and Neural Network

故障特征提取和识别方法的研究对发展和完善电力电子装置的智能故障诊断技术有着重要的作用。应用小波包能量法提取出电力电子装置在各种状态下电压及电流信号的能量特征向量,并将它们进行数据融合作为神经元网络故障分类器的输入向量,由神经元网络故障分类器对各种故障进行识别和诊断。以电力电子整流装置主电路故障为例进行了仿真实验,试验结果表明该方法无需数学模型就能快速准确的完成故障定位诊断。

The research on failure feature extraction and failure identification method plays an important role in developing and improving the intelligent fault diagnostic. These signal feature vectors of various states about voltage and current of power electronic device, which are extracted by using the wavelet packet energy method, are integrated as the input vectors of the neural network failure classifier, identified and diagnosed various faults by the neural network classifier. The result of the simulate experiment with the main circuit of electronic power electronic rectifier device failure as an example, shows that the method can deal with the location of fault diagnosis rapidly and accurately without mathematical model.