山东电力技术2026，Vol.53Issue(5)：83-92,10.DOI:10.20097/j.cnki.issn1007-9904.250193
基于变分模态分解与对抗多层感知相结合的铁塔局部螺栓松动分析
Analysis of Local Bolt Loosening in Iron Towers Based on the Combination of Variational Mode Decomposition and Mahalanobis Distance
摘要
Abstract
The bolts of transmission towers play a vital role in ensuring the structural stability and operation safety of the towers.Therefore,accurate and efficient bolt looseness detection technology is particularly important for evaluating the condition of these structures.In this paper,a method combining variational modal decomposition(VMD)and against multilayer perceptual networks(AT-MLP)is proposed to identify the looseness of transmission tower bolts.The acoustic sensor collects the voiceprint signal of the iron tower bolts,which is then decomposed using VMD to extract characteristic information,such as the center frequency of the voiceprint signal associated with loose bolts.The time-frequency information of the voiceprint is input into the MLP model as a characteristic parameter to identify the state of the bolts.The results show that the recognition accuracy for bolt looseness of 110 kV and 220 kV transmission towers is 88%and 80%,respectively,while the recognition accuracy for the normal state of the bolts is 100%and 98%,respectively.These results meet the actual operation and maintenance requirements of towers.The method proposed in this paper can accurately distinguish fastening bolts from loose bolts,which provides a basis for the state evaluation of transmission towers.关键词
输电铁塔/螺栓松动/对抗训练/多层感知器/声学Key words
transmission tower/loose bolts/confrontation training/multilayer perceptron/acoustics分类
机械制造引用本文复制引用
秦勇,姚建光,李捷,宗鹏鹏,周燠..基于变分模态分解与对抗多层感知相结合的铁塔局部螺栓松动分析[J].山东电力技术,2026,53(5):83-92,10.基金项目
国网江苏省电力有限公司科技项目"输电铁塔螺栓松动声纹检测技术研究与智能检测装置研制"(J2023156).Science and Technology Project of State Grid Jiangsu Electric Power Com-pany"Research on Detection Technology of Bolt Looseness Voiceprint of Transmission Tower and Development of Intelligent Detection Device"(J2023156). （J2023156）
