基于分形神经网络的电力电子电路故障诊断

Fault diagnosis of power electronic circuits based on fractal and neural network

针对电力电子电路的故障,分析了故障产生的特征类型,提出了基于分形理论及BP网络故障诊断的方法。以三相整流桥路为例,利用分形理论建立了故障元与分形维数之间的关系,对故障信息做预处理。通过仿真试验提取出用于BP神经网络训练的学习样本,并构建了用于不同类故障的三层BP神经网络结构,继而确定故障点。

The feature types of fault occurred in power electronic circuit are analyzed. The methods of fractal theory and back propagation （BP） neural network are proposed. In three-phase rectifier bridge circuit, the relationship between the fault element and the fractal dimension of fault information are set up to do preprocessing. The learning samples for training BP neural network is obtained by simulating the different fault types of thyristors in the rectifier, and the three-layer BP neural networks for different fault types are constructed. Then the point of fault is also determined.