摘要
针对传统的轴承故障诊断过于依赖专家经验和故障特征提取困难的现状,同时为了适应故障诊断的大数据处理及实时监测的需求,提出了一种基于变分模态分解(variational mode decomposition,VMD)与发育神经网络(developmental neural network,DNN)相结合的故障诊断方法。先将原始信号分组处理,再对分组后的信号进行VMD分解,得到若干个模态分量(IMF),根据相关系数对信号进行重构,随后提取重构信号的各个模态分量(IMF)的能量占比组成特征向量组,输入发育神经网络中进行训练和测试,进而对故障类型进行识别与分类,并与支持向量机(SVM)进行了对比。实验表明该方法的识别准确率可高达98.3%。
Aiming at the current situation that traditional bearing fault diagnosis is too dependent on expert experience and the difficulty of fault feature extraction,and in order to adapt to the needs of fault diagnosis big data processing and real-time monitoring,a variational mode decomposition(variational mode decomposition,VMD)based on Fault diagnosis method combined with developmental neural network(developmental neural network,DNN).First,the original signal is grouped,and then the grouped signal is subjected to VMD decomposition to obtain several modal components(IMF).The signal is reconstructed according to the correlation coefficient,and then the modal components(IMF)of the reconstructed signal are extracted The energy proportion forms a feature vector group,which is input into the developmental neural network for training and testing,and then the fault type is identified and classified,and compared with the support vector machine(SVM).Experiments show that the recognition accuracy of this method can be as high as 98.3%.
作者
赵志杰
曾劲松
郝旺身
司少剑
ZHAO Zhi-jie;ZENG Jin-song;HAO Wang-shen;SI Shao-jian(School of Mechanical and Power Engineering,Zhengzhou University,Zhengzhou 450001,China)
出处
《组合机床与自动化加工技术》
北大核心
2021年第5期81-85,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
国家重点研发计划项目(2016YFF0203100)
河南省教育厅科学技术研究重点项目(13A460673)
河南省教育厅自然科学研究项目(2011B460012)。
关键词
滚动轴承
变分模态分解
发育神经网络
特征向量
故障检测
rolling bearing
variational modal decomposition
developmental neural network
eigenvectors
fault detection
