摘要
针对机电作动器的传统故障诊断方法依赖于人工特征提取和工程经验的问题,该文提出一种基于一维卷积神经网络(1DCNN)的智能故障诊断方法。相较于传统故障诊断算法中特征提取和分类的分开处理,该方法将两者合二为一、共同进行。首先,利用重叠采样对直驱型机电作动器的正常信号和故障信号进行预处理来获取数据样本;然后将样本输入到设计的一维卷积神经网络模型中,通过多层数据变换得到有效的特征表示,从而建立原始数据端与运行状态端之间的映射关系,实现机电作动器端到端的故障诊断。实验结果表明,该方法可以有效地诊断出机电作动器的故障,且故障识别率可以达到98%左右。另外,该方法在不同白噪声下仍可以保持较高的故障识别率,具有比较好的鲁棒性和泛化能力。
To address the problem that traditional fault diagnosis methods of electromechanical actuators largely depend on artificial feature extraction and engineering experience,this paper proposes an intelligent fault diagnosis method based on one dimensional convolutional neural network(1DCNN).Compared with the separation of feature extraction and classification in the traditional fault diagnosis algorithm,the proposed method combines the two into one.Firstly,the normal signals and fault signals of direct-driven electromechanical actuators are preprocessed by overlapping sampling to acquire data samples.Subsequently,the obtained samples are fed into the designed one-dimensional convolutional neural network model,and the effective feature representation is acquired through multi-layer data transformation,thereby establishing a mapping relationship between the raw data and operating state and achieving end-to-end fault diagnosisof electromechanical actuators.The experimental results demonstrate that the proposed algorithm can effectively diagnose the fault of the electromechanical actuator,and the fault recognition accuracy can reach about 98%.In addition,the proposed method can still maintain a high fault recognition accuracy under different white noise conditions,which shows that it has good robustness and generalization performance.
作者
李世晓
杜锦华
龙云
Li Shixiao;Du Jinhua;Long Yun(State Key Laboratory of Electrical Insulation and Power Equipment Xi’an Jiaotong University,Xi’an 710049 China)
出处
《电工技术学报》
EI
CSCD
北大核心
2022年第S01期62-73,共12页
Transactions of China Electrotechnical Society
基金
国家自然科学基金项目(51877172)
中央高校基本科研业务费专项资金项目(1191329824)
陕西省自然科学基础研究计划青年项目(2019JQ-458)资助。
关键词
直驱型机电作动器
一维卷积神经网络
故障诊断
深度学习
Direct-drive electromechanical actuator
one-dimensional convolutional neural network
fault diagnosis
deep learning
