基于时间序列的系统盲辨识与特征提取  

Blind System Identification and Feature Extraction Based on Time Series

作　　者：张锐[1] 黄晋英[1] 郎忠宝 

机构地区：[1]中北大学机械与动力工程学院,太原030051 [2]晋西集团技术中心,太原030051

出　　处：《机械工程师》2015年第3期35-36,共2页Mechanical Engineer

摘　　要：分析了基于时间序列的盲系统辨识方法与特征提取方法,并且对A R M A模型系统盲辨识进行了仿真,仿真结果表明基于累积量的参数确定方法优于B urg法。建立了齿轮箱A R M A系统盲辨识模型,提取时间序列模型的系数作为故障诊断的特征向量,证明了结构发生故障时系统特性的变化与故障之间具有一定的对应关系。This paper analyzes the method of blind system identification and characteristics extraction based on time series,simulates blind identification of the ARMA model system. The simulation results show that the parameter determination method based on cumulative amount is superior to Burg. The authors establish the blind identification model of a gearbox ARMA system,extract time series model coefficient as the feature vector of fault diagnosis,prove that there is certain corresponding relationship between the changes of the system characteristics and fault when the structure failure occurs.

关 键 词：时间序列 盲辨识 特征提取 齿轮箱 

分 类 号：TH133[机械工程—机械制造及自动化] TN911[电子电信—通信与信息系统]