摘要
齿轮泵动态压力信号蕴含了丰富的状态信息,是齿轮泵的关键性能参数之一;负熵是随机变量独立性的自然测度,反映了机械信号信息的动态变化特征。为了进行基于齿轮泵动态压力信号的故障诊断,在引入经验模态分解技术的基础上,提出了齿轮泵压力信号的负熵基ICA特征提取方法,并进而联合最小二乘支持向量机,探讨了基于ICA和LS-SVM的齿轮泵特征提取和故障诊断方法;理论和试验研究表明,基于负熵的ICA特征提取和故障诊断方法是有效的。
Gear pumps＇ dynamic pressure signals contain abundant condition information.They are key parameters of gear pumps.Negentropy is a natural measurement for independent property of random variables.It denotes dynamic information features of mechanical signals.To study gear pumps＇ fault diagnosis based on their dynamic pressure signals,ICA feature abstraction algorithm based on negentropy and EMD was proposed.It combined excellent characteristics of negentropy and empirical mode decomposition.After negentropy ICA features abstracted,least squares support vector machines was applied to recognize different fault patterns.Their applications in gear pumps indicate that ICA feature abstraction and fault diagnosis algorithm based on negentropy is effective.
出处
《流体机械》
CSCD
北大核心
2011年第6期5-9,24,共6页
Fluid Machinery
基金
总装备部预研重点基金项目
关键词
压力信号
负熵
经验模态分解
最小二乘支持向量机
特征提取
故障诊断
pressure signal
negentropy
empirical mode decomposition
least squares support vector machine
feature abstraction
fault diagnosis
