海上风电机组塔架微弱振动信号自动化检测  

Automated detection of weak vibration signals in offshore wind turbine tower structures

作　　者：张险峰 高亚州 雒德宏 马璐 傅广泽 ZHANG Xianfeng;GAO Yazhou;LUO Dehong;MA Lu;FU Guangze(China Three Gorges Corporation.,Wuhan 430014,China)

机构地区：[1]中国长江三峡集团有限公司,武汉430014

出　　处：《自动化与仪器仪表》2025年第1期35-38,共4页Automation & Instrumentation

基　　金：国家重点研发计划资助:大型海上风电机组测试与性能提升关键技术及应用示范(2023YFB4203100)。

摘　　要：海风、海浪、水汽等因素会对风电机组塔架产生振动和噪声,导致风电机组塔架区域的微弱振动检测精度下降。对此,提出海上风电机组塔架微弱振动信号自动化检测技术。通过粒子群算法确定最佳小波阈值函数。采用优化后的小波阈值函数对塔架结构的振动信号进行去噪。利用去噪后的正常振动信号和微弱振动信号的时频谱图来构建自监督特征提取器,并选择轻量级卷积网络Mobilenet V2对提取器进行训练,以实现对微弱振动信号的提取和检测。实验结果表明:所提方法可以精准检测出信号中存在的微弱振动,检测时间最长仅为12.52 ms,具有实用性。Factors such as sea breeze,waves,and water vapor can cause vibration and noise in the wind turbine tower,leading to a decrease in the accuracy of weak vibration detection in the wind turbine tower area.In this regard,an automated detection technology for weak vibration signals of offshore wind turbine towers is proposed.Determine the optimal wavelet threshold function through particle swarm optimization algorithm.Using optimized wavelet threshold functions to denoise the vibration signals of tower structures.Using the time-frequency spectrum of denoised normal vibration signals and weak vibration signals to construct a self supervised feature extractor,and selecting the lightweight convolutional network Mobilenet V2 for training the extractor to achieve the extraction and detection of weak vibration signals.The experimental results show that the proposed method can accurately detect weak vibrations in the signal,with a detection time of only 12.52ms at the longest,indicating practicality.

关 键 词：海上风电机组 小波去燥 自监督特征提取器 微弱振动 自动化检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]