摘要
为了提高齿轮泵行星轮的典型故障诊断精度,提出了一种基于经验模态分解(EEMD)和双向长短时记忆网络(Bi-LSTM)的行星齿轮泵故障诊断方法。研究结果表明:通过模型精度和耗时的最优参数为节点数200和网络层数4层。本网络损失小于1%,满足良好稳定性的条件,可以实现精确识别齿面磨损和缺齿故障,断齿、正常齿的轮识别率都达到了93%以上,齿根裂纹故障识别率达到了86.5%。对信号EEMD分解后,可以促进BiLTSM模型所有分量都获得更优的时序性,促使模型诊断精度得到显著提升。Bi-LTSM模型到达后期迭代过程时,可以更快拟合,获得高于LTSM的验证精度。该研究对提高机械传动设备的故障识别能力,具有一定的理论指导意义。
In order to improve the typical fault diagnosis accuracy of gear pump planetary gear,a new method is proposed based on ensemble empirical mode decomposition(EEMD)and bi-directional long short time memory network(Bi-LSTM)planetary gear pump fault diagnosis method.The results show that the optimal parameters of model accuracy and time consumption are as follows:the number of nodes is 200 and the number of network layers is four.The loss of the network is less than1%,which meets the condition of good stability,and can accurately identify tooth surface wear and missing tooth faults.The wheel recognition rate of broken tooth and normal tooth is more than 93%,and the tooth root crack fault is 86.5%.The EEMD decomposition of signals can promote the better timing of all components of the Bi-LTSM model and significantly improve the diagnostic accuracy of the model.When the Bi-LTSM model reaches the later iteration process,it can be fitted faster and obtain higher verification accuracy than LTSM.The research has certain theoretical significance to improve the fault identification ability of mechanical transmission equipment.
作者
高美真
高烨童
GAO Meizhen;GAO Yetong(School of Information Engineering,Jiaozuo Normal College,Jiaozuo 454000,Henan,China;School of Computer and Information Engineering,Xi’an University of Technology,Xi’an 710061,Shaanxi,China)
出处
《中国工程机械学报》
北大核心
2022年第3期263-268,共6页
Chinese Journal of Construction Machinery
基金
国家自然科学基金资助项目(61170014)。
关键词
齿轮泵
故障诊断
经验模态分解
双向长短时记忆网络
分类精度
gear pump
fault diagnosis
ensemble empirical mode decomposition
bi-directional long short-time memory
classification accuracy
