基于形态学广义分形维数的电力电子电路故障诊断  

Fault diagnosis of power electronic circuits based on mathematical morphology based generalized fractal dimensions

作　　者：宋平岗[1] 周军[1] 

机构地区：[1]华东交通大学电气与电子工程学院,江西南昌330013

出　　处：《电工电能新技术》2014年第12期53-59,共7页Advanced Technology of Electrical Engineering and Energy

摘　　要：电力电子电路故障检测和诊断,是保证电力电子装置安全、可靠运行的一个重要手段。针对传统电力电子电路故障诊断方法存在的局限性,提出了一种基于数学形态学广义分形维数的新方法。结合分形几何理论,求取不同故障的输出电压波形的分形维数,以此作为故障特征,再与神经网络相结合,进行有效识别。以一个实际的Buck功率电路为例,对其各类故障信号进行了分析。结果表明,与传统的盒计数法计算的广义分形维数相比,形态学广义分形维数能够更加有效地区分Buck电路在不同状态下的信号,并且数学形态学只涉及简单的加减和取大、取小运算,因此计算简单快速,估计准确稳定,为准确判断电力电子电路故障诊断提供了一种快速有效的新方法。The fault detection and diagnosis of power electronic circuits can guarantee safe operation of power electronic devices.Considering the limitation of traditional fault diagnosis of power electronic circuits,this paper presents a new method based on mathematical morphology based generalized fractal dimensions.Combined with the fractal geometry theory,this method extracts the fractal dimensions of different faults＇ output waveforms as the fault characteristics,and identifies them by Elman neural network.This paper takes a Buck circuit as the example and analyzes its fault signals.The results show that compared with the traditional box-counting based generalized fractal dimensions,the mathematical morphology based generalized fractal dimensions can distinguish Buck circuit signals in different states more effectively.Besides,the calculation of mathematical morphology is simple and accurate for it only involves addition,subtraction practice,and maximum and minimum operations.Thus,it provides a fast and effective method for accurate fault diagnosis of power electronic circuits.

关 键 词：数学形态学 多重分形 广义分形维数 电力电子电路 故障诊断 

分 类 号：TM461[电气工程—电器] TP183[自动化与计算机技术—控制理论与控制工程]