基于数学形态学广义分形维数逆变器故障诊断  

Inverter Fault Diagnosis Based on Generalized Fractal Dimensions of Mathematical Morphology

作　　者：宋平岗[1] 章伟[1] 林家通 游小辉 罗剑[1] 

机构地区：[1]华东交通大学电气学院,江西南昌330013

出　　处：《华东交通大学学报》2016年第6期110-117,共8页Journal of East China Jiaotong University

基　　金：国家自然科学基金(51367008)

摘　　要：基于逆变器开路故障输出电流波形的差异,提出将形态学广义分形维数运用于逆变器故障检测中。将逆变器三相输出电流按a,b,c三相依次取六个不同参数下的形态学广义分形维数作为ELM神经网络的输入,以故障类型作为神经网络的输出。仿真结果显示该方法的故障区分率高达97.62%。基于三相输出电流的形态广义分形维数能够准确的识别出逆变器在各种故障状态下的电流信号,为逆变器开路故障诊断提供了一种简单准确的新方法。Based on the current waveform differences output by inverters with open-circuit fault, this paper puts forward the application of mathematical morphology-based generalized fractal dimensions for inverter fault diag- nosis. Mathematical morphology-based generalized fractal dimensions with six parameters in the sequence of three-phase output current（a,b,c） are taken as the input of Elman Neural Network, and fault types are set as the output. The simulation result shows that the fault discrimination rate is up to 97.62%. The current signals under various conditions of breakdown can be identified properly by mathematical morphology-based generalized fractal dimensions of three-phase output current, which may provide a new and precise method for open＇circuit fault diagnosis.

关 键 词：逆变器 故障诊断 数学形态学 广义分形维数 

分 类 号：TM464[电气工程—电器]