低转速航空发动机滚动轴承故障深度异常检测方法  

Deep anomaly detection method for low-rotating speed rolling bearing faults of aero-engine

作　　者：康玉祥 陈果[2] 盛嘉玖 王浩 尉询楷[3] KANG Yuxiang;CHEN Guo;SHENG Jiajiu;WANG Hao;WEI Xunkai(College of Civil Aviation,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China;College of General Aviation and Flight,Nanjing University of Aeronautics and Astronautics,Liyang 213300,China;Beijing Aeronautical Engineering Technology Research Center,Beijing 100076,China)

机构地区：[1]南京航空航天大学民航学院,南京210016 [2]南京航空航天大学通用航空与飞行学院,江苏溧阳213300 [3]北京航空工程技术研究中心,北京100076

出　　处：《振动与冲击》2024年第7期186-195,共10页Journal of Vibration and Shock

基　　金：国家科技重大专项(J2019-IV-004-0071);国家自然科学基金项目(52272436)。

摘　　要：针对航空发动机滚动轴承在低转速状态下故障难检测的问题,提出了一种基于Transformer框架的深度支持向量描述方法用于检测低转速滚动轴承的故障。首先,构建了基于Transformer模型的振动特征提取主干网络。然后,将所提取的特征输入一个三层自编码器结构,用于计算网络模型的损失函数。为减少网络计算量,提高训练速度,在预处理中将滚动轴承的振动加速度时域信号通过快速傅里叶变换(FFT)得到的频谱结果作为网络的输入,且仅依靠正常数据完成模型的训练。最后,在带机匣的航空发动机转子试验器和某型真实的航空发动机上分别进行了试验验证。结果表明,所提方法能够准确的实现对低转速滚动轴承故障的检测,且检测精度分别为93%和100%,充分表明该方法具有很好的异常检测能力及应用价值。Here,aiming at the problem of aero-engine rolling bearing faults at low rotating speed being difficult to detect,a deep support vector description method based on Transformer framework was proposed to detect faults of low rotating speed rolling bearings.Firstly,a vibration feature extraction backbone network based on Transformer model was constructed.Then,the extracted features were input into a 3-layer autoencoder structure to calculate the loss function of network model.In order to reduce network computation amount and improve training speed,time-domain vibration acceleration signals of rolling bearing were pre-processed and the frequency spectrum results obtained using fast Fourier transform(FFT)were taken as the input of network to complete training of model only using normal data.Finally,test verifications were performed on rotor tester of an aero-engine with casing,and a real aero-engine of a certain type,respectively.The results showed that the proposed method can correctly detect faults in low rotating speed rolling bearing with detection accuracies of 93% and 100%,respectively;the proposed method can have excellent anomaly detection ability and application value.

关 键 词：低转速 滚动轴承 深度异常检测 TRANSFORMER 航空发动机 

分 类 号：TG156[金属学及工艺—热处理]