首页 | 本学科首页   官方微博 | 高级检索  
     

基于未知输入观测器的非线性时间序列故障预报
引用本文:张正道,胡寿松.基于未知输入观测器的非线性时间序列故障预报[J].控制与决策,2005,20(7):769-772.
作者姓名:张正道  胡寿松
作者单位:1. 南京航空航天大学,自动化学院,南京,210016;江南大学,控制理论与工程研究中心,江苏,无锡,214122
2. 南京航空航天大学,自动化学院,南京,210016
基金项目:国家自然科学基金重点项目(60234010),国防基础科研项目(K1603060318),航空科学基金项目(02E52025).
摘    要:将未知非线性系统的输出作为时间序列并进行空间重构,针对得到的离散线性时变系统,提出了基于未知输入观测器的预测新方法.以实时拟合时间序列的线性AR模型作为时变系统的已知线性部分,将拟合误差作为时变系统的未知输入,实现了对非线性时间序列的一步预测.再利用递推预测的方法,将一步预测推广到N步预测,同时证明了该方法的预测误差有界.通过未知输入的预测值和状态的预测误差的变化可以方便地判断故障的发生,实现故障预报.仿真结果证明了方法的有效性.

关 键 词:故障预报  非线性时间序列  未知输入观测器
文章编号:1001-0920(2005)07-0769-04

Fault Prediction for Nonlinear Time Series Based on Unknown Input Observer
ZHANG Zheng-dao,HU Shou-song.Fault Prediction for Nonlinear Time Series Based on Unknown Input Observer[J].Control and Decision,2005,20(7):769-772.
Authors:ZHANG Zheng-dao  HU Shou-song
Affiliation:ZHANG Zheng-dao~
Abstract:The nonlinear time series, which is formed by the output of the unknown system, is converted into discrete time-varying dynamic system with space reconstruction. A novel method is presented for the prediction of time series which is achieved by the adaptive observation of system states. An AR model is used to approximate the linear part of the time series; the nonlinear part and the approximate error are regarded as the unknown-input of system. A one-step-ahead prediction method is proposed. The one-step-ahead prediction is then extended to N-steps-ahead moving horizon prediction, and the prediction error is proved bounded. The fault is predicted conveniently by the prediction of unknown-input and prediction error. The simulation results show the method is efficient.
Keywords:Fault prediction  Nonlinear time series  Unknown input observe
本文献已被 CNKI 维普 万方数据 等数据库收录!
点击此处可从《控制与决策》浏览原始摘要信息
点击此处可从《控制与决策》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号