摘要
现有针对行星齿轮箱的故障诊断方法一般仅研究单一故障,但实际行星齿轮箱的故障一般由多个故障耦合而成,耦合故障的故障机理比单一故障的故障机理更复杂,振动信号中的非线性因素对特征提取的干扰更严重。针对该问题,提出了一种基于精细复合多尺度散度熵(RCMDE)、等距特征映射(ISOMAP)和遗传算法优化核极限学习机(GA-KELM)的行星齿轮箱耦合故障诊断方法。首先,利用振动加速度计采集了行星齿轮箱单一故障和耦合故障下运行时的振动信号,构建了故障数据集;随后,利用RCMDE提取了行星齿轮箱振动信号的故障特征,建立了初始的特征样本;接着,利用ISOMAP对故障特征进行了降维,并以可视化的方式获取了低维的特征样本;最后,将新特征输入至GA-KELM分类器中,对行星齿轮箱的不同故障类型进行了识别,并基于行星齿轮箱多点损伤样本,对RCMDE方法的可靠性进行了研究。研究结果表明:基于RCMDE和ISOMAP的故障特征提取方法能够有效提取振动信号中的故障特征,而GA-KELM的故障诊断准确率达到了98.13%,平均诊断准确率达到了96.25%。相较其他故障特征提取方法,基于RCMDE、ISOMAP和GA-KELM的行星齿轮箱耦合故障诊断方法能够更好地诊断行星齿轮箱的耦合故障,具有更高的诊断准确率。
The existing fault diagnosis methods for planetary gearbox generally only studied a single fault,but the actual planetary gearbox fault was usually caused by the coupling of multiple faults.The fault mechanism of coupling fault was more complicated than that of a single fault,and the nonlinear factors in vibration signal had more serious interference to feature extraction.To solve this problem,a planetary gearbox coupling fault diagnosis method based on refined composite multiscale diversity entropy(RCMDE),isometric feature mapping(ISOMAP),and genetic algorithm optimized kernel extreme learning machine(GA-KELM) was proposed.Firstly,vibration accelerometer was used to collect vibration signals of planetary gearbox under single fault and coupling fault,and the fault data set was constructed.Secondly,RCMDE was used to extract the fault features of the planetary gearbox vibration signal and establish initial feature samples.Then,ISOMAP was used to reduce the dimension of fault features and visualize them to obtain low-dimensional feature samples.Finally,the new features were input into the GA-KELM classifier to realize the identification of different fault types of planetary gearbox.The reliability of RCMDE method was studied based on multi-point damage samples of planetary gearbox.The research results show that the fault feature extraction method based on RCMDE and ISOMAP can effectively extract the fault feature from the vibration signal,while the fault diagnosis accuracy of GA-KELM is 98.13%,and the average diagnosis accuracy is 96.25%.Comparing with other fault feature extraction methods,the proposed method can diagnose the coupling fault of planetary gearbox better and has higher diagnostic accuracy.
作者
苏世卿
王华锋
SU Shiqing;WANG Huafeng(School of Numerical Control Technology,Xinxiang Vocational and Technical College,Xinxiang 453006,China;School of Mechanical Automation,Wuhan University of Science and Technology,Wuhan 430081,China)
出处
《机电工程》
CAS
北大核心
2024年第9期1584-1594,共11页
Journal of Mechanical & Electrical Engineering
基金
国家自然科学基金资助项目(52075395)。
关键词
齿轮传动
耦合故障
故障诊断准确率
精细复合多尺度散度熵
等距特征映射
遗传算法优化核极限学习机
gear drive
coupling fault
fault diagnosis accuracy
refined composite multiscale diversity entropy(RCMDE)
isometric feature mapping(ISOMAP)
genetic algorithm optimized kernel extreme learning machine(GA-KELM)
