Two-stage parameter estimation algorithms for Box–Jenkins systems

• Published in: IET signal processing Vol. 7; no. 8; pp. 646 - 654
• Main Authors: Ding, Feng, Duan, Honghong
• Format: Journal Article
• Language: English
• Published: Stevenage The Institution of Engineering and Technology 01.10.2013; Institution of Engineering and Technology; John Wiley & Sons, Inc
• Subjects: Algorithms; Applied sciences; BJ system decomposition; Box‐Jenkins systems; Computer simulation; Detection, estimation, filtering, equalization, prediction; Estimates; Exact sciences and technology; gradient methods; Information, signal and communications theory; least squares approximations; Least squares method; Mathematical models; Noise; noise model parameter algorithm; parameter estimation; Recursive; recursive estimation; Signal and communications theory; Signal processing; Signal, noise; stochastic processes; system model parameter algorithm; Telecommunications and information theory; two‐stage multiinnovation stochastic gradient method; two‐stage parameter estimation algorithms; two‐stage recursive least‐square identification method; least squares approximations; system model parameter algorithm; signal processing; Box-Jenkins systems; two-stage parameter estimation algorithms; two-stage recursive least-square identification method; noise model parameter algorithm; two-stage multiinnovation stochastic gradient method; BJ system decomposition; recursive estimation; parameter estimation; stochastic processes; gradient methods; Parameter estimation; Innovation; Multistage method; Simulation; Least squares method; Subsystem; Recursive method; Stochastic method; Algorithm; Gradient method
• ISSN: 1751-9675; 1751-9683; 1751-9683
• DOI: 10.1049/iet-spr.2012.0183

A two-stage recursive least-squares identification method and a two-stage multi-innovation stochastic gradient method are derived for Box–Jenkins (BJ) systems. The key is to decompose a BJ system into two subsystems, one containing the parameters of the system model and the other containing the parameters of the noise model, and then to estimate the parameters of the system model and the noise model, respectively. The simulation examples indicate that the proposed algorithms can generate highly accurate parameter estimates and require small computational burden.