Skip to main navigation Skip to search Skip to main content

Iterative learning identification with bias compensation for Stochastic linear time-varying systems

  • Fazhi Song
  • , Yang Liu*
  • , Zhile Yang
  • , Xiaofeng Yang
  • , Ping He
  • *Corresponding author for this work
  • Fudan University
  • Harbin Institute of Technology

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

A novel iterative learning algorithm is proposed for the identification of linear time-varying (LTV) output-error (OE) systems that perform tasks repetitively over a finite-time interval. Conventional LTV system identification normally relies on recursion algorithms in time domain, which are unable to follow fast changing parameters because of an inevitable estimation lag. To overcome this problem, an extra iteration axis is introduced besides the time axis in the parameter estimation process, and identification algorithm performed in iteration domain is proposed. Firstly, a norm-optimal identification approach is presented to balance the tradeoff between convergence speed and noise robustness. Then a bias compensation algorithm is further proposed to improve the estimation accuracy. Finally, numerical examples are provided to validate the algorithm and confirm its effectiveness. The algorithm is effective to estimate both slow and abrupt parameter changes with high accuracy without estimation lags.

Original languageEnglish
Title of host publicationIntelligent Computing, Networked Control, and Their Engineering Applications - International Conference on Life System Modeling and Simulation, LSMS 2017 and International Conference on Intelligent Computing for Sustainable Energy and Environment, ICSEE 2017, Proceedings
EditorsDong Yue, Tengfei Zhang, Chen Peng, Dajun Du, Min Zheng, Qinglong Han
PublisherSpringer Verlag
Pages231-239
Number of pages9
ISBN (Print)9789811063725
DOIs
StatePublished - 2017
EventInternational Conference on Life System Modeling and Simulation, LSMS 2017 and International Conference on Intelligent Computing for Sustainable Energy and Environment, ICSEE 2017 - Nanjing, China
Duration: 22 Sep 201724 Sep 2017

Publication series

NameCommunications in Computer and Information Science
Volume762
ISSN (Print)1865-0929

Conference

ConferenceInternational Conference on Life System Modeling and Simulation, LSMS 2017 and International Conference on Intelligent Computing for Sustainable Energy and Environment, ICSEE 2017
Country/TerritoryChina
CityNanjing
Period22/09/1724/09/17

Keywords

  • Bias compensation
  • Linear time-varying systems
  • Output error systems
  • System identification

Fingerprint

Dive into the research topics of 'Iterative learning identification with bias compensation for Stochastic linear time-varying systems'. Together they form a unique fingerprint.

Cite this