基于PCA和BP网络的液压油缸内泄漏故障诊断  

Internal leakage fault diagnosis of hydraulic cylinder using PCA and BP network

作　　者：唐宏宾[1,2] 吴运新[1] 滑广军[1] 马昌训[1] 

机构地区：[1]中南大学机电工程学院,湖南长沙410083 [2]长沙理工大学汽车与机械工程学院,湖南长沙410004

出　　处：《中南大学学报（自然科学版）》2011年第12期3709-3714,共6页Journal of Central South University:Science and Technology

基　　金：国家高技术研究发展计划("863"计划)项目(2008AA042801;2008AA042802)

摘　　要：针对液压油缸内泄漏故障诊断时,由于提取的时域参数过多,导致诊断速度慢、实时性差等问题,提出基于主成分分析(PCA)和BP神经网络的诊断方法。该诊断方法是:首先提取压力信号的时域参数作为初始特征,然后利用PCA方法将高维初始特征空间压缩到低维最终特征空间,并将得到的最终特征输入到BP神经网络进行故障模式识别。研究结果表明:该诊断方法在满足故障检测识别率的同时提高了诊断速度,为液压油缸内泄漏的故障诊断提供了一种实用方法。Based on the fact that in fault diagnosis for internal leakage of hydraulic cylinder,many time-domain features are selected from pressure singal to make diagnosis slow and unreal-time,a fault diagnosis approach based on principal component analysis（PCA） and BP network was proposed.According to the method,time-domain paramers of pressure singal were selected as prime features at first,and then PCA was used to get the final features.At last,these final features were input into BP network to identify faults.The results show that the method can meet the recognition rate and increase diagnosis speed,so it can be used in the leakage fault diagnosis of hydraulic cylinder.

关 键 词：液压油缸内泄漏 故障诊断 主成分分析 BP网络 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]