时频特征联合VGG-16的风力发电机齿轮箱故障诊断方法  

Fault Diagnosis Method for Wind Generator Gearbox with Time-Frequency Feature Combined with VGG-16

作　　者：于大海[1] 郝俊红[1] 高严 YU Dahai;HAO Junhong;GAO Yan(School of Electrical Information,Changchun Guanghua University,Changchun,Jillin 130011,China)

机构地区：[1]长春光华学院电气信息学院,吉林长春130011

出　　处：《自动化应用》2023年第17期72-74,78,共4页Automation Application

摘　　要：针对传统风力发电机齿轮箱故障诊断存在效率低、实时性较差、准确率偏低等问题,本文提出了一种时频特征联合深度学习的风力发电机齿轮箱故障诊断方法。通过实时测取的箱体振动信号,利用离散小波转换提取信号的时频特征,并联合改进的VGG-16模型,完成齿轮组不同类型故障的诊断。实验结果表明,所设计诊断方法的Recall和mAP值较高,分别为93.51%、91.85%;断齿、磨损以及根裂故障的诊断准确率均在90%以上,且检测实时性良好,能较好地满足实际应用需求。To solve the problems of low efficiency,poor real-time performance and low accuracy of traditional wind turbine gearbox fault diagnosis,this paper proposes a time-frequency feature combined with in-depth learning method for wind turbine gearbox fault diagnosis.By using the real-time measured vibration signal of the box,the time-frequency characteristics of the signal are extracted by the discrete wavelet transformation,and the improved VGG-16 model is combined to complete the diagnosis of different types of faults of the gear set.The results show that the Recall and mAP values of the designed diagnostic formula are 93.51%and 91.85%,respectively.The diagnostic accuracy of broken teeth,wear and root cracks is more than 90%,and the real-time detection is good,which can better meet the practical application requirements.

关 键 词：时频特征 改进 VGG-16 离散小波 故障诊断 

分 类 号：TP274.421[自动化与计算机技术—检测技术与自动化装置]