基于SDAE和GRUNN的行星齿轮故障识别  

Fault identification of planetary gears based on the SDAE and GRUNN

作　　者：于军 高莲莲[4] 于广滨[5] 刘可 郭振宇 YU Jun;GAO Lianlian;YU Gangbin;LIU Ke;GUO Zhenyu(Key Laboratory of Advanced Manufacturing and Intelligent Technology,Harbin University of Science and Technology,Harbin 150080,China;State Key Laboratory of Process Automation in Mining&Metallurgy,Beijing 100089,China;School of Automation,Harbin University of Science and Technology,Harbin 150080,China;College of Electrical and Electronic Engineering,Harbin University of Science and Technology,Harbin 150080,China;School of Mechanical and Power Engineering,Harbin University of Science and Technology,Harbin 150080,China)

机构地区：[1]哈尔滨理工大学先进制造智能化技术教育部重点实验室,哈尔滨150080 [2]矿冶过程自动控制技术国家重点实验室,北京100089 [3]哈尔滨理工大学自动化学院,哈尔滨150080 [4]哈尔滨理工大学电气与电子工程学院,哈尔滨150080 [5]哈尔滨理工大学机械动力工程学院,哈尔滨150080

出　　处：《振动与冲击》2021年第2期156-163,共8页Journal of Vibration and Shock

基　　金：矿冶过程自动控制技术国家重点实验室开放基金(BGRIMM-KZSKL-2020-06);国家自然科学基金(61806060);黑龙江省杰出青年基金(JC2015013)。

摘　　要：针对噪声环境和时变转速工况下行星齿轮故障识别率低的问题,提出一种基于堆叠消噪自动编码器(SDAE)和门控循环单元神经网络(GRUNN)的行星齿轮故障识别方法。构建基于SDAE和GRUNN的混合模型,处理前后关联的时序数据,自动地从含噪样本中提取鲁棒故障特征;将行星齿轮故障诊断的训练样本看作该混合模型的输入数据,采用Adam优化算法和dropout技术训练该混合模型,实现多参数的优化,防止过拟合现象的发生;根据训练后的混合模型,利用softmax分类器识别待诊样本中行星齿轮的状态。通过行星齿轮的故障识别实验验证该方法的有效性,实验结果表明该方法具有较强的抗噪能力和时变转速适应能力。In order to address the problem of low fault identification accuracy of planetary gears under noisy environment and time-varying rotational speed conditions,a fault diagnosis method for planetary gears using the stacked denoising autoencoder(SDAE)and gated recurrent unit neural network(GRUNN)was proposed.A hybrid model based on the SDAE and GRUNN was constructed to process pre and post correlation time-series data,and automatically extract robust fault features.The training samples for planetary gear fault diagnosis were regarded as the input data of the hybrid model.The Adam optimization algorithm and the dropout technique were employed to train the hybrid model so as to realize the optimization of multiple parameters and prevent from overfitting.A softmax classifier was employed to identify the planetary gear states of test samples according to the hybrid model after training.The effectiveness of the proposed method was validated through a fault identification experiment of planetary gears.The experimental results demonstrate that the proposed method is of stronger anti-noise ability and excellent adaptability to time-varying rotational speed.

关 键 词：行星齿轮 故障识别 噪声环境 时变转速 堆叠消噪自动编码器(SDAE) 门控循环单元神经网络(GRUNN) 

分 类 号：TH17[机械工程—机械制造及自动化] TP18[自动化与计算机技术—控制理论与控制工程]