基于BP神经网络对减速箱齿轮副的故障诊断与仿真研究  

Fault diagnosis and simulation research for gearbox gear based on BP neural networks

作　　者：贺艳[1] 李培菊[2] 邹兆东[3] 马文涛[1] 

机构地区：[1]新疆工程学院机械工程系,新疆乌鲁木齐830091 [2]新疆油田公司陆梁作业区,新疆克拉玛依834000 [3]新疆油田公司石西作业区,新疆克拉玛依834000

出　　处：《机械研究与应用》2012年第6期170-172,174,共4页Mechanical Research & Application

摘　　要：齿轮故障诊断的基本方法是利用振动信号,通过提取特征信息,比较各次测量中的差异进行诊断。根据BP神经网络诊断方法,运用虚拟样机技术建立齿轮模型,模拟各种各种故障,提取振动信号,然后给出BP神经网络训练的特征频率,使其对数据进行分析学习。当输入待诊断样本后,BP网络对数据进行对比分析得出诊断结果。经过仿真检验表明该方法对减速箱齿轮副的故障诊断有效,对其他旋转机械的故障诊断和维修保养也具有一定的实用价值。The gear fault diagnosis method is based on the vibration signal,through extracting the feature information,comparing the measurements difference diagnose.Based on the BP neural network diagnosis method using virtual prototype technique and gear model is built,simulating various fault,the vibration signals are extracted;and then the training of BP neural network characteristic frequency is given,the data is analyzed and studied.When the diagnostic samples are input,BP network is compared and analyzed for data,and the diagnosis results are got.Through the simulation test,it shows that the method is effective for gear fault diagnosis in the reduction gear box,and also has a certain practical value for other rotating machinery fault diagnosis,repair and maintenance.

关 键 词：BP神经网络 时域平均分析法 齿轮副故障诊断 仿真分析 

分 类 号：TH132[机械工程—机械制造及自动化]