基于LMD-SECNN的风机变桨系统故障检测
Fault detection of wind turbine pitch system based on LMD and attention mechanism
井露茜 1文传博1
作者信息
- 1. 上海电机学院 电气学院, 上海 201306
- 折叠
摘要
风机变桨系统控制较为复杂,且易发生机械故障.针对振动信号比较复杂,直接利用神经网络进行学习会影响特征的提取,提出了一种以将局部均值方法及注意力机制相结合的故障诊断模型,即LMD-SECNN模型.首先利用局部均值分解法对数据进行预处理,减少噪声干扰,并最大限度保存原始信号里的故障特征;其次经过LMD处理后产生多个PF分量;最后以神经网络模型Inception v1架构为基础进行改进,增加通道注意力SEnet模块.实验结果表明:LMD-SECNN模型的准确率达到99.42%,远高于对比模型的准确率,验证了所提方法的有效性和优越性.
Abstract
The main part of the fan pitch system is the pitch gearbox,which has multiple moving parts,and its control is more complex and prone to mechanical failure In order to solve the problem that the direct use of neural network learning will affect the extraction of features due to the complexity of vibration signals,and the fault data is much less than the normal data,a fault diagnosis model combining the local mean method and the attention mechanism is proposed,namely the LMD-SECNN model.Firstly,the local mean decomposition method is used to preprocess the data,which reduces the noise interference and preserves the fault characteristics in the original signal to the greatest extent.After LMD processing,multiple PF components are generated,which also solves the problem of less fault data to a certain extent.The neural network model Inception v1 architecture is improved on the basis,and the channel attention SEnet module is added,which not only reduces the number of parameters,but also avoids the problem of excessive model computation,making the fault characteristics more obvious.Experimental verification shows that the accuracy of the LMD-SECNN model reaches 99.42%,which is much higher than that of the comparison model,which verifies the effectiveness and superiority of the proposed method.
关键词
风机变桨系统/局部均值分解/注意力机制/故障检测Key words
wind turbine pitch system/Local mean decomposition/attention mechanisms/fault detection引用本文复制引用
出版年
2024
NETL
NSTL
万方数据