基于小波包分形的瓦斯传感器故障诊断方法  

Gas sensor fault diagnosis method based on wavelet package fractal analysis

作　　者：陈宏[1] 邓芳明[2] 吴翔[2] 付智辉[2] 

机构地区：[1]华东交通大学轨道交通学院,江西南昌330013 [2]华东交通大学电气与自动化工程学院,江西南昌330013

出　　处：《传感器与微系统》2016年第11期26-29,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(21265006);江西省科技支撑计划资助项目(20161BBE50076;20161BBE50077)

摘　　要：针对瓦斯传感器的故障诊断问题,提出一种基于小波包分形的瓦斯传感器故障诊断方法。使用3层小波包对故障信号进行分解和重构,获得不同频带的重构信号,计算各个重构信号的分形维度,并构成对应的故障特征向量。以此作为输入向量来训练支持向量机(SVM),完成故障的诊断。实验结果表明:该方法能有效地提取传感器的故障特征,提高了传感器故障诊断的准确率,可有效地应用于瓦斯传感器的故障诊断。Aiming at fault diagnosis problem of gas sensor,a gas sensor fault diagnosis method based on wavelet package fractal analysis is proposed. Fault signals are decomposed and reconstructed by using three-level wavelet package,reconstructed signals of different frequency bands are achieved. Compute fractal dimension of each reconstructed signal,and compose corresponding fault feature vectors. Inputting these fault vectors to train SVM to achieve fault diagnose. Experimental result shows that the proposed method extract effectively features of fault of sensor and increase of fault diagnosis,which can be applied to fault diagnosis of gas sensor effectively.

关 键 词：瓦斯传感器故障诊断 小波包变换 分形分析 支持向量机 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]