基于HSMM的机械故障演化预测诊断研究  

Study on Prediction and Diagnosis of Mechanical Fault Evolution Based on HSMM Model

作　　者：于宁[1] 王艳红[1] 蔡明[2] 田中大[1] 

机构地区：[1]沈阳工业大学信息科学与工程学院,沈阳110870 [2]东北大学机械工程与自动化学院,沈阳110819

出　　处：《组合机床与自动化加工技术》2018年第1期56-59,共4页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：国家自然科学基金资助项目(51375082);中央高校基本科研业务专项资金资助项目(N160306001)

摘　　要：为了给机械设备提供更准确的故障预测诊断,采用小波分析的方法对滚动轴承的振动信号进行特征提取与分析,并提出一种新混合模型(即将状态空间模型与隐半马尔可夫模型相结合的混合模型)的故障预测诊断方法。首先在动态观测系统中建立故障状态方程,将故障作为关键因子,并在混合后的模型中给予相应的证明,通过对其分析处理、使用预测模型进行训练以及对比分析设备的退化状态,给出合理的预测方案,然后对其进行深入分析,最后得出研究结论。In order to provide more accurate fault prediction and diagnosis of mechanical equipment,the wavelet analysis method was used to extract and analyze the vibration signal of rolling bearing,and a newhybrid model of fault prediction and diagnosis method was proposed,which combined the state space model with the hidden semi-Markov model. Firstly,the fault state equation was established in the dynamic observation system,the fault was taken as the key factor,and the corresponding proof was given in the hybrid model. Through analyzing and processing of it,training of the prediction model,comparing and analyzing of the equipment degradation state,the reasonable prediction scheme was given. Then deeply analysis was conducted. Finally,the conclusions were drawn.

关 键 词：故障预测 滚动轴承 小波分析 残差 状态识别 

分 类 号：TH122[机械工程—机械设计及理论] TG506[金属学及工艺—金属切削加工及机床]