基于迁移学习的复杂环境下输电杆塔异常振动识别  

Abnormal Vibration Identification of Transmission Line Tower Based on Transfer Learning

作　　者：张恒[1] 徐广辉 饶丹[1] 李临风 周华良[1] ZHANG Heng;XU Guanghui;RAO Dan;LI Linfeng;ZHOU Hualiang(NARI Technology Co.,Ltd.,Nanjing 211106,China)

机构地区：[1]国电南瑞科技股份有限公司,江苏南京211106

出　　处：《电力信息与通信技术》2022年第1期61-67,共7页Electric Power Information and Communication Technology

摘　　要：架空输电线路杆塔所处环境复杂,传统方法中假设的标准样本与实际样本分布一致的前提遭到破坏,导致单一的识别模型在不同环境下对杆塔异常振动识别准确率降低。为改善识别模型偏差问题,文章提出一种基于领域适配深度迁移学习的杆塔异常振动识别方法。通过一维卷积神经网络实现不同环境条件下异常振动信号的自动特征提取,并引入迁移学习,实现复杂环境下异常振动的准确识别。所提方法通过优化场景差异性正则化损失函数,缩小复杂真实场景与典型场景间的分布差异,获得有效的领域适配模型。实验结果表明,所提方法能明显改善复杂环境下输电杆塔异常振动识别效果,并提高识别准确率。The environment of overhead transmission line tower is complex,and the premise that the distribution of actual samples is consistent with that of standard samples in traditional method is destroyed.This situation results in the low identification accuracy of abnormal vibration for single identification model under different conditions.In order to improve the identification model bias,this paper proposes an abnormal vibration identification method based on the deep transfer learning for domain adaptation.The one-dimensional convolutional neural network is are utilized to auto extract the features for the abnormal vibration signal under different working conditions.Then,the transfer learning is introduced to achieve the accurate identification of tower abnormal vibration under complex conditions.The proposed method reduces the distribution difference between complex real scenes and typical scenes through optimizing scene difference regularized loss function,and obtains an effective domain adaptation model.The experiment results show that the proposed method can obviously improve the recognition results of overhead transmission line tower abnormal vibration under complex conditions,and enhance the identification accuracy.

关 键 词：杆塔振动识别 迁移学习 卷积神经网络 领域适配 

分 类 号：TM74[电气工程—电力系统及自动化]