摘要
主元分析(PCA)是一种利用数据之间相依性建立系统低维模型的方法,该方法将高度相关的过程数据投影到低维空间,并保留原有的有用信息。本文将主元分析方法引入到液压系统的泄漏检测中,与传统的故障检测方法相比,主元分析方法采用HotellingT2和Q统计作为故障检测的依据,具有不依赖过程数学模型的特点。基于多维数据驱动的PCA方法,建立液压泄漏监测装置,并将该装置应用到宝钢二期煤焦推焦机液压系统当中,结果表明泄漏监测系统能够及时、准确地发现系统泄漏故障。
Principal component analysis,which projects sampling data to low-dimension spaces and preserves useful information,is a method to establish low-dimension models based on mutual dependency of sampling data.The Hotelling T^2 and Q statistics were used as rules for fault diagnosis in PCA-based method.Compared with model-based approach,it is simple and straight-forward.Based on data-driven PCA method,an oil leakage monitoring device was developed.The good performance of the leakage monitoring device was verified by the application of door-drawing hydraulic system in Baosteel coke pusher No.2.The results show the oil leakage fault can be detected timely and rightly.
出处
《机床与液压》
北大核心
2008年第4期199-201,共3页
Machine Tool & Hydraulics
基金
国家自然科学基金资助项目(50575200)
关键词
主元分析
故障监测
取门液压系统
油液泄漏
Principal component analysis
Fault detection
Door-drawing hydraulic system
Oil leakage