基于马田系统的设备健康监测技术研究

提出了一种基于马田系统的设备健康检测与故障分类方法,利用设备运行的振动信号作为信号源进行了验证;首先介绍了马田系统的实施步骤,提出了基于马田系统的设备故障诊断与分类方案;其次,分别利用信号小波变换模极大值估计得到的Lipschitz指数和Hilbert-Huang变换进行特征提取;最后利用凯斯西储大学电气工程实验室的轴承振动数据对方法进行了验证,采用滚动轴承的5种状态振动信号:正常,滚动体故障(轻微、严重),内圈故障(轻微、严重),基于信号奇异性特征的故障检测率为100%,故障分类率为81.8%,基于信号能量特征的故障检测率和故障分类率均为100%;结果表明了方法的正确性,利用Hilbert-Huang变换提取特征进行诊断和分类的正确率更高。

Research on Equipment Health Monitoring Using Mahalanobis-Taguchi System

In this paper,an Mahalanobis-Taguchi system based health monitoring & fault diagnosis and classification scheme is presented,vibration signals are used as the signal resource.Firstly,the procedure of implementing Mahalanobis-Taguchi System is introduced,a fault diagnosis and classification method is proposed based-on MTS.Secondly,Lipschitz Exponents and Hilbert-Huang transform are used to extract characteristic vectors.Lastly,the method is validated by Case Western Reserve University rotating machinery seeded-fault-test vibration signals vibration signal of five states are used,that is normal,ball defect(slight,defect),inner race defect(slight,defect),fault detection rate is 100% and classification rate 81.8% using signal singularity characteristics,compared to 100% and 100% using energy characteristics.Results showed that this method performed well,the feature extraction method using Hilbert-Huang transform has higher accuracy.