基于小波包和支持向量机的舰用发动机故障诊断  

Fault Diagnosis of Certain Key Components Based on Wavelet Packet and SVM Warship Engine

作　　者：崔建国[1] 刘宝胜[1] 王桂华[2] 于明月[1] 高阳[2] 

机构地区：[1]沈阳航空航天大学自动化学院,沈阳110136 [2]沈阳发动机设计研究所,沈阳110000

出　　处：《火力与指挥控制》2016年第6期181-184,共4页Fire Control & Command Control

基　　金：辽宁省自然科学基金(2014024003);航空科学基金(2010ZD54012);国防预研基金(A0520110023);国防基础科研基金资助项目(Z052012B002)

摘　　要：舰用发动机是一个复杂的大系统,由于受到海洋恶劣气候的影响,其故障发生的概率大大增加,因此,对舰用发动机故障诊断进行研究具有重大实际意义。以舰用发动机的主泵轴承为例,提出了基于小波包和支持向量机的故障诊断方法。首先采用振动加速度传感器获取轴承的振动信号,然后对采集数据进行多层小波包分解,求各频带信号能量,形成各种故障模式下的特征向量。将形成的故障特征向量训练集输入到支持向量机,通过训练建立诊断分类器,并运用测试数据对建立的诊断分类器进行测试。实验结果表明,该方法可以很好地实现舰用发动机故障诊断效能,具有很好的工程应用前景。Warship engine is a complex system,due to the impact of marine climate,the probability of failure is greatly increased,and therefore,the warship engine fault diagnosis research is of great significance. In this paper,taking the main pump bearing of the ship engine as an example,the fault diagnosis method based on wavelet packet and support vector machine is presented. First,the collected data multilayer wavelet packet and find the signal energy of each band,formed feature vector various failure modes. Fault feature vectors formed training set input to the support vector machine,to establish the diagnosis classifier through training and the use of test data established diagnostic classification test. The experimental results show that the method can well realize the ship diesel engine fault diagnosis efficiency and has good prospects for engineering applications.

关 键 词：支持向量机 小波包 故障诊断 特征向量 

分 类 号：TP211.2[自动化与计算机技术—检测技术与自动化装置]