基于大模型混合算法的变电站缺陷检测方法研究  

Research on Substation Defect Detection Method Based on Large Model Hybrid Algorithm

作　　者：童超 TONG Chao

机构地区：[1]浙江千宇物联科技有限公司,浙江杭州310051

出　　处：《电力系统装备》2025年第1期140-142,共3页Electric Power System Equipment

摘　　要：为提高变电站运行的可靠性,实现对其缺陷的精准检测,引进大模型混合算法,以某变电站为例,开展缺陷检测方法的设计.对采样信号进行转换,通过信号处理技术,进行变电站设备运行中PD信号的分离;利用大模型混合算法中的随机森林决策树,进行变电站信号特征的提取;引进阈值判断法,进行变电站设备状态感知与缺陷检测.对比试验结果表明,设计方法检测结果中分离的信号幅值较高,可以直接通过此方法判定变电站设备存在缺陷,且检测结果与实际结果一致.In order to improve the reliability of substation operation and achieve accurate detection of its defects,a large model hybrid algorithm is introduced.Taking a certain substation as an example,a defect detection method is designed.Sampling signals are used to separate PD signals during the operation of substation equipment through signal processing techniques;Extracting substation signal features using the random forest decision tree in the large model hybrid algorithm;Introduce threshold judgment method for substation equipment status perception and defect detection.The comparative experimental results show that the amplitude of the separated signals in the detection results of the designed method is relatively high,which can directly determine the existence of defects in substation equipment through this method,and the detection results are consistent with the actual results.

关 键 词：大模型混合算法 PD信号 分离 检测方法 

分 类 号：TM63[电气工程—电力系统及自动化]