变压器振动信号分析在重载铁路数字孪生变电所的应用研究

Application of Transformer Vibration Signal Analysis in Digital Twin Substation for Heavy-Haul Railway

基于重载铁路数字孪生变电所建设,对变压器振动信号分析应用开展了重点研究。首先,采用经验小波变换对变压器振动信号进行分解处理,获取反映变压器振动特性的多阶经验小波分量;其次,以精细化复合多尺度分散熵RCMDE作为衡量各阶经验小波分量时域分布特性的指标,通过组合各阶经验小波分量对应的RCMDE,表征变压器振动信号的多维度特征;最后,采用优化的OPTICS聚类算法辨识变压器振动是否存在异常,优化的OPTICS算法能够有效改善建模样本不平衡的问题,使得构建的预警模型整体性能更为稳健。在朔黄铁路变压器安装测试设备,采集分析振动信号,对研究方法进行了验证。经过实例验证,该研究方法适用于变压器故障在线监测,对于辨识变压器机械性故障预警具有一定实用价值。

Based on the construction of digital twin substations for heavy-haul railways,key research has been conducted on the analysis and application of transformer vibration signals.Firstly,empirical wavelet transform was employed to decompose and process the transformer vibration signals for obtaining multi-order empirical wavelet components that reflect the transformer vibration characteristics.Secondly,the refined composite multi-scale dispersed entropy RCMDE was adopted as an indicator to measure the temporal distribution characteristics of each order of empirical wavelet components.By combining the RCMDE corresponding to each order of empirical wavelet components,the multi-dimensional characteristics of transformer vibration signals were characterized.Finally,the optimized OPTICS clustering algorithm was utilized to identify whether there were abnormalities in transformer vibration.This algorithm can improve the problem of imbalanced modeling samples,making the overall performance of the built early warning model more robust.Meanwhile,this paper collected and analyzed vibration signals,and validated the research method by installing testing equipment on transformers of the Shuohuang Railway.Practical verification shows that this method is suitable for online monitoring of transformer faults and has certain practical significance for identifying mechanical fault warning of transformers.