小波包与改进BP神经网络相结合的齿轮箱故障识别  

Fault Diagnosis of Gearbox Using Wavelet Package and Improved BP Neural Network

作　　者：时建峰[1,2] 程珩[1] 许征程[2] 史少辉[3] 时伟[4] 钮效鵾 

机构地区：[1]太原理工大学机械电子工程研究所,太原030024 [2]石家庄学院国有资产管理处,石家庄050035 [3]石家庄学院物理系,石家庄050035 [4]石家庄学院科研处,石家庄050035 [5]<石家庄学院学报>编辑部,石家庄050035

出　　处：《振动．测试与诊断》2009年第3期321-324,共4页Journal of Vibration,Measurement & Diagnosis

基　　金：山西省科技攻关项目(编号:2007032054)

摘　　要：应用小波包分解技术提取齿轮箱振动信号中的故障特征向量,并以此作为改进BP神经网络的输入,对神经网络进行训练,建立了齿轮箱运行状态分类器,用以识别齿轮箱的运行状态。试验结果表明,小波包分解与神经网络相结合的齿轮箱齿轮故障识别方法是可靠的,可以准确识别齿轮箱的故障。The wavelet package was applied to decompose vibration signals of a gearbox to get the fault feature vectors. The feature vectors were employed as the input samples to train an improved BP neural network,and then the running state classifier of the gearbox fault was set up. The experimental results show that the proposed method is effective for gearbox fault diagnosis.

关 键 词：小波包 BP神经网络 齿轮箱 故障识别 

分 类 号：TH133.33[机械工程—机械制造及自动化]