基于Park-HHT的故障特征提取方法  

Fault feature extraction method based on Park-HHT

作　　者：李波 张琳[1] 汪文峰[1] 王冠男 LI Bo;ZHANG Lin;WANG Wenfeng;WANG Guannan(Air Defense and Anti-Missile College,Air Force Engineering University,Xi’an 710043,China;Unit 93786 of the PLA,Zhangjiakou 075061,China;Unit 93114 of the PLA,Beijing 100195,China)

机构地区：[1]空军工程大学防空反导学院,陕西西安710043 [2]中国人民解放军93786部队,河北张家口075061 [3]中国人民解放军93114部队,北京100195

出　　处：《系统工程与电子技术》2020年第12期2944-2952,共9页Systems Engineering and Electronics

基　　金：中国博士后科学基金(2017M623417)资助课题。

摘　　要：针对三相对称系统,提出了一种新的故障特征提取方法。该方法首先将系统输出信号进行Park变换,之后进行希尔伯特-黄变换(Hilbert-Huang transform,HHT),从HHT的本征模态函数中选取低频分量,提取信号幅度作为故障特征值。该方法兼具Park变换和HHT的优点,在降低变量维度的同时提高了故障分辨率,并且具有较强的抗噪声能力。最后,在三相逆变电路中进行了仿真实验,根据特征向量间距离分布对不同方法的提取效果进行评价,验证了所提方法的有效性。A new fault feat ure extraction method is proposed for three-phase symmetric system.Firstly,the three-phase output signal of the system is firstly Park transformed by the method,and then the signal goes through Hilbert-Huang transform(HHT).The low-frequency component is selected from the eigenmode function of HHT and the signal amplitude is extracted as the fault characteristic value.This method has the advantages of both Park transformation and HHT,which can improve the fault resolution while reducing the variable dimension,and has a strong anti-noise capability.Finally,the simulation experiment is carried out in the three-phase inverter circuit,and the extraction effect of different methods is evaluated by using the distance distribution between feature vectors,which verify the effectiveness of the proposed method.

关 键 词：故障诊断 特征提取 PARK变换 希尔伯特-黄变换 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]