Title :
Weighted least squares algorithm for continuous-time model
Author_Institution :
Dept. of Math., Kansas Univ., Lawrence, KS, USA
Abstract :
In this paper we discuss a weighted least squares algorithm for the following continuous-time model: A(S)yt=SB(S)ut+C(S)vt where S denotes the integral operator, i.e. Syt=∫0tysds and A(S), B(S) and C(S) are matrix polynomials in the integral operator S. Similar to discrete WLS, the almost sure boundedness and convergence are established for continuous-time weighted least squares (CWLS) algorithm. The simulation results for the discrete WLS and the discrete extended least squares (ELS), the CWLS and the continuous ELS algorithms are given in this paper. Our techniques of proofs are totally different from those in discrete case. The results presented can be extended to state-space models with continuous time
Keywords :
autoregressive moving average processes; continuous time systems; convergence of numerical methods; identification; least squares approximations; almost sure boundedness; continuous ARMAX model; continuous-time model; continuous-time weighted least squares; convergence; discrete extended least squares; identification; integral operator; matrix polynomials; state-space models; weighted least squares algorithm; Algorithm design and analysis; Convergence; Integral equations; Least squares methods; Mathematical model; Mathematics; Parameter estimation; Polynomials; Stochastic processes; Stochastic systems;
Conference_Titel :
Decision and Control, 1994., Proceedings of the 33rd IEEE Conference on
Conference_Location :
Lake Buena Vista, FL
Print_ISBN :
0-7803-1968-0
DOI :
10.1109/CDC.1994.410920
