@article{oai:metro-cit.repo.nii.ac.jp:00000129,
 author = {青木, 立 and Aoki, Tatsu and 川田, 誠一 and Kawata, Seiichi},
 journal = {東京都立産業技術高等専門学校研究紀要, Research reports of Tokyo Metropolitan College of Industrial Technology},
 month = {Mar},
 note = {P(論文), The mathematical model of a plant is required in order to design a control system. Especially, the system parametersmust be estimated in real-time to keep the desired control performance in the case of the plant parameter variations.Though the number of identification method are proposed, they need much amount of calculation and the calculationsmethods are based on double precision floating point arithmetic. Then, it is difficult for the fixed point microprocessorwhose word length is short to calculate the algorithm precisely within sampling interval. Thus, the aim of this research isto propose a system identification method for the embedded mechatronic control systems. As a first step of this research,system parameter estimation on 2nd-order system is considered. ARX model is adopted as a system model and systemparameters are estimated by the recursive least squares with forgetting factor. As a disturbance, noise is injected in theplant output, dead zone and offset are injected in the plant input to verify the robustness of the algorithm. Simulationresults show that the measurement noise on the plant output gives most large effect at the parameter estimation error.},
 pages = {19--23},
 title = {システム同定アルゴリズムに関するロバスト性の検証-逐次最小二乗法に関する検証-},
 volume = {6},
 year = {2012},
 yomi = {アオキ, タツ and カワタ, セイイチ}
}