基于轻量化Transformer模型的风机叶片质量不平衡故障诊断  

Diagnosis of Wind Turbine Blade Mass Imbalance Based on a Lightweight Transformer Model

作　　者：李京明 成庶[1] 向超群[1] 刘红文 蒋韬 周瑞睿 LI Jingming;CHENG Shu;XIANG Chaoqun;LIU Hongwen;JIANG Tao;ZHOU Ruirui(School of Traffic&Transportation Engineering,Central South University,Changsha,Hunan 410075,China;CRRC Zhuzhou Institute Co.,Ltd.,Zhuzhou,Hunan 412001,China)

机构地区：[1]中南大学交通运输工程学院,湖南长沙410075 [2]中车株洲电力机车研究所有限公司,湖南株洲412001

出　　处：《控制与信息技术》2025年第1期13-20,共8页CONTROL AND INFORMATION TECHNOLOGY

基　　金：国家重点研发计划项目(2022YFB4201602-02)。

摘　　要：风机叶片质量不平衡会导致风机在旋转时产生较大的离心力,进而引发整个系统的机械振动,长期处于振动情况下运行,风机叶片、主轴和传动系统等关键部件会产生应力集中,缩短风机使用寿命及其发电效率。针对风机叶片不平衡造成的风电机组振动信号难以提取以及模型诊断精度低、模型结构复杂的问题,文章提出一种基于变分模态分解(variational mode decomposition,VMD)信号输入的轻量化Transformer模型,用于风机叶片不平衡故障诊断。首先对原始的振动信号进行变分模态分解,将其分解为多个模态的时域信号,增大故障信号的维度;其次,为了降低模型的复杂度,提出一种不含位置编码的改进Transformer诊断模型,简化模型复杂度;最后通过监测和数据采集(supervisory control and data acquisition,SCADA)系统采集风机机舱振动数据进行实验分析。实验结果表明,所提方案可以实现对风机叶片质量不平衡故障诊断,其精度可达到98.3%。Mass imbalance in wind turbine blades can result in significant centrifugal forces during rotation,leading to mechanical vibrations throughout the turbine system.Prolonged operation under vibration conditions can cause stress concentration in key components such as the blades,main shaft,and the transmission system,ultimately reducing the equipment's lifespan and lowering the power generation efficiency.To address the challenges of signal extraction for vibrations induced by blade imbalance,as well as the low diagnostic accuracy and complex structures of existing models,this paper proposes a lightweight Transformer model based on signal input subjected to variational mode decomposition(VMD)for diagnosing mass imbalance in wind turbine blades.Original vibration signals are decomposed into multi-modal time-domain signals to increase the dimensionality of fault signals.To reduce the model's complexity,an improved Transformer diagnostic model without positional encoding is introduced,thereby simplifying the model structure.An experimental analysis was conducted using a supervisory control and data acquisition(SCADA)system to collect nacelle vibration data.Based on the experimental results,the proposed approach achieved a diagnostic accuracy of up to 98.3% for mass imbalance in wind turbine blades.

关 键 词：风电机组 变分模态分解 轻量化 Transformer模型 位置编码 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置] TM315[自动化与计算机技术—控制科学与工程]