Title :
Modified least-squares identification of linear systems with noisy input and output observations
Author_Institution :
Dept. of Math., Univ. of Western Sydney, Kingswood, NSW, Australia
Abstract :
In this paper a new type of bias-eliminated least-squares (BELS) based algorithm is proposed for consistent identification of linear systems with noisy input and output measurements. It is shown that estimation of the noise variances can be implemented when the degree of the denominator polynomial of the system transfer function is increased by one. The modified BELS algorithm is attractive and meaningful in that noisy data are used in identification with no prefiltering and a direct estimate of system parameters is given without any parameter transformation
Keywords :
least squares approximations; linear systems; noise; parameter estimation; transfer functions; bias-eliminated least-squares based algorithm; denominator polynomial; linear systems; modified least-squares identification; noise variances; noisy input observations; noisy output observations; transfer function; Australia; Linear systems; Mathematics; Measurement standards; Noise measurement; Noise robustness; Parameter estimation; Pollution measurement; Polynomials; Transfer functions;
Conference_Titel :
Decision and Control, 1996., Proceedings of the 35th IEEE Conference on
Conference_Location :
Kobe
Print_ISBN :
0-7803-3590-2
DOI :
10.1109/CDC.1996.574640
