基于图谱功率谱熵和最大均值差异的GIS机械状态辨识方法

Mechanical State Identification Method of GIS Based on Spectral Power Spectral Entropy and Maximum Mean Difference

针对常规方法对于气体绝缘金属封闭开关设备(Gas Insulated Switchgear,GIS)机械缺陷的特征识别稳定性差、识别率低的问题,在图谱理论的基础上,提出一种基于图谱功率谱熵和最大均值差异(Maximum Mean Discrepancy,MMD)的GIS机械状态辨识方法。首先将采集得到的GIS振动信号转化为图信号,并利用图傅里叶变换技术变换至图谱域进行分析处理;然后提取图谱功率谱熵作为表征GIS不同状态的特征参数;最后利用MMD距离判别函数实现GIS不同工况下的状态辨识。实验结果表明:在噪声干扰的情况下,所提方法能够有效提取GIS不同状态下的特征参数,并成功区分出屏蔽罩松动及内部异物缺陷,状态辨识精度高达93.89%,较常规方法有明显提高。

Aiming at the problems of poor feature recognition stability and low recognition rate of mechanical defects in gas insulated switchgear by conventional methods,based on graph spectrum theory,a new mechanical state identification of GIS based on spectrum power spectrum entropy and maximum mean discrepancy is proposed.Firstly,the collected GIS vibration signal are converted into graph signal,and then converted into graph domain by graphic Fourier transform technology for analysis and processing;Then the graph power spectrum entropy is extracted as a characteristic parameter to characterize different states of GIS;Finally,the MMD distance discriminant function is used to realize the status identification of GIS under different working conditions.The experimental results show that the proposed method can effectively extract the characteristic parameters of GIS in different states under the condition of noise interference,and successfully distinguish the loose shield from the internal foreign body defects.The accuracy rate of state recognition is as high as 93.89%,which is significantly higher than that of the conventional method.