基于一维密集连接卷积网络的风电齿轮箱智能故障分类

人工智能技术的飞速发展为现代能源装备的精益化故障诊断与健康管理提供了可能。风电齿轮箱由多个齿轮、轴承组成,且长期在变速、变载荷工况下运行,依靠传统的故障特征提取结合机器学习方法进行故障诊断存在精度低、缺乏智能性等缺点。文章提出了基于一维密集连接卷积网络的风电齿轮箱故障分类方法:将原始振动信号直接送入网络模型,经过密集连接、合成连接与卷积运算,匹配对应的故障类型,迭代训练故障分类模型;振动信号输入模型后的分类结果决定所属故障类别。文章所提出的风电齿轮箱故障分类方法具有诊断流程简单、故障识别率高等特点,多工况试验台故障数据验证了该方法的有效性。

Intelligent fault classification of wind turbine gearbox based on 1-D densely connected convolutional network

The development of artificial intelligence provides a platform for accurate fault diagnosis and health management for energy equipment.Wind turbine gearbox consists of multiply gears and bearings,and operates under varying speeds and loads.For the fault diagnosis of wind turbine gearbox,conventional fault feature extraction combined with machine learning have some disadvantages,e.g.,low accuracy and intelligence insufficiency,etc.In this paper,a 1-D densely connected convolutional network based fault classification method is proposed for wind turbine gearbox.The original vibration signals are directly input into the designed model,and processed by the dense connectivity,composite function and convolutional computation.Through matching the fault types,the model is iteratively trained.During the testing phase,the classification results from original testing signals will determine the final fault types.The proposed method is simple in fault diagnosis and high accuracy in fault recognition,which is verified by the experimental vibration data of multiply operational conditions and faults.