基于SDP和MCNN-LSTM的齿轮箱故障诊断方法  

Gearbox fault diagnosis based on SDP and MCNN-LSTM

作　　者：吴胜利[1] 周燚 邢文婷 WU Shengli;ZHOU Yi;XING Wenting(School of Traffic and Transportation,Chongqing Jiaotong University,Chongqing 400074,China;School of Management Science and Engineering,Chongqing Technology and Business University,Chongqing 400067,China)

机构地区：[1]重庆交通大学交通运输学院,重庆400074 [2]重庆工商大学管理科学与工程学院,重庆400067

出　　处：《振动与冲击》2024年第15期126-132,178,共8页Journal of Vibration and Shock

基　　金：国家自然科学基金项目(51705052);重庆市自然科学基金项目(cstc2019jcyj-msxmX0779);国家社会科学基金项目(17CGL003)。

摘　　要：齿轮箱在长期使用过程中,不可避免地会产生齿轮故障和轴承故障,严重影响传动精度和设备运行安全。基于此,针对齿轮箱常见故障类型,研究多通道对称点图案(symmetrized dot pattern, SDP)数据处理方法,并利用最小能量误差法实现SDP关键参数的选取。结合多尺度卷积神经网络(multi-scale convolutional neural network, MCNN)的空间处理优势、长短时记忆网络(long short term memory, LSTM)的时间处理优势及其良好的抗噪性和鲁棒性,提出了一种基于SDP和MCNN-LSTM的齿轮箱故障诊断模型。同时利用东南大学齿轮箱数据集,验证了基于SDP和MCNN-LSTM的齿轮箱故障诊断方法对齿轮和轴承常见故障类型特征提取的有效性,并与现有其他故障诊断方法进行对比,结果表明了所提方法具有更高的精度。During long-term use,gearbox inevitably experiences gear and bearing failures which seriously affect transmission accuracy and equipment operation safety.Here,aiming at common types of faults in gearbox,a multi-channel symmetrical dot pattern(SDP)data processing method was studied,and the minimum energy error method was used to select key parameters of SDP.Combining spatial processing advantage of multi-scale convolutional neural network(MCNN)and temporal processing advantage of long short-term memory(LSTM)as well as its good noise resistance and robustness,a gearbox fault diagnosis model based on SDP and MCNN-LSTM was proposed.Meanwhile,the gearbox dataset of Southeast University was used to verify the effectiveness of the proposed gearbox fault diagnosis method based on SDP and MCNN-LSTM in extracting features of common fault types of gears and bearings.This method was compared with other existing fault diagnosis methods.The results showed that the proposed method has higher accuracy.

关 键 词：齿轮箱故障诊断 对称点图案(SDP) 最小能量误差 多尺度卷积神经网络(MCNN) 长短时记忆网络(LSTM) 

分 类 号：TH113.1[机械工程—机械设计及理论]