[关键词]
[摘要]
为了有效诊断气体绝缘金属封闭输电线路(GIL)的机械故障,搭建了110 kV GIL试验平台并设计了3种典型机械故障,通过互补集合经验模态分解(CEEMD)模糊熵值与鲸鱼优化极限学习机(WOA-ELM)模型联合方法对GIL机械故障模式进行识别与诊断。首先,利用CEEMD方法对振动信号进行分解,引入正负白噪声组对信号进行处理,得到含有故障信息的模态分量(IMF)。其次,利用模糊熵计算模态分量特征值,得到能表征故障特征的模糊熵值。最后,结合WOA-ELM模型对特征向量集进行模式识别,根据聚类结果与自适应阈值对GIL设备机械故障进行诊断和预警。结果表明,利用CEEMD与模糊熵对GIL振动信号特征进行分析,可以有效避免模态混叠和冗余噪声分量的干扰,得到能够表征故障特征的特征值;利用WOA-ELM模型可以有效实现GIL设备机械故障诊断与预警。
[Key word]
[Abstract]
In order to diagnosis mechanical faults in gas insulated transmission line (GIL) effectively, the 110 kV GIL experimental platform is built and three typical mechanical faults are designed. The joint method of complementary ensemble empirical mode decomposition (CEEMD) fuzzy entropy and whale optimization algorithm-extreme learning machine (WOA-ELM) model is used to identify and diagnose the GIL mechanical fault mode. Firstly, the CEEMD method is used to decompose the vibration signal, and the positive and negative white noise is introduced to process the signal to obtain the modal component with fault information. Secondly, the fuzzy entropy is used to calculate the eigenvalues of modal component to obtain the fuzzy entropy which can represent the fault features. Finally, the WOA-ELM model is used to recognize the eigenvector set, and the clustering results and adaptive threshold are used to analyze the fault features. The mechanical fault of GIL equipment can be diagnosed and forewarned by using the value. The results show that, combining CEEMD and fuzzy entropy to analyze GIL vibration signal can effectively avoid the interference of modal aliasing and redundant noise components, and obtain the eigenvalues that can characterize the fault characteristics; using WOA-ELM model can effectively realize GIL equipment mechanical fault diagnosis and early warning.
[中图分类号]
[基金项目]
中国博士后科学基金项目(2020M671318);中央高校基本科研任务经费项目(B200202173);江苏省自然科学基金青年基金项目(BK20190490);国网江苏省电力公司重点科技项目(J2020040)