Computers and Mathematics with Applications 60 (2010) 1200–1208
Contents lists available at ScienceDirect
Computers and Mathematics with Applications
journal homepage: www.elsevier.com/locate/camwa
Gradient-based iterative parameter estimation for Box–Jenkins systems
I
Dongqing Wang
a
, Guowei Yang
a
, Ruifeng Ding
b,∗
a
College of Automation Engineering, Qingdao University, Qingdao 266071, PR China
b
School of Communication and Control Engineering, Jiangnan University, Wuxi 214122, PR China
a r t i c l e i n f o
Article history:
Received 20 March 2009
Received in revised form 31 May 2010
Accepted 1 June 2010
Keywords:
Signal processing
Recursive dentification
Parameter estimation
Stochastic gradient
Iterative algorithms
Box–Jenkins models
a b s t r a c t
This paper presents a gradient-based iterative identification algorithms for Box–Jenkins
systems with finite measurement input/output data. Compared with the pseudo-linear
regression stochastic gradient approach, the proposed algorithm updates the parameter
estimation using all the available data at each iterative computation (at each iteration),
and thus can produce highly accurate parameter estimation. An example is given.
© 2010 Elsevier Ltd. All rights reserved.
1. Introduction
The least-squares and stochastic gradient parameter estimation methods are two classes of basic identification algo-
rithms. They have received much attention in many areas, e.g., signal processing, adaptive control and system identifi-
cation [1–9]. These two methods are used in studying different types of systems, e.g., multivariable systems [2,10–14],
dual-rate and multirate sampled-data systems [14–18], nonlinear block-oriented systems [19–22], and the performances of
these two classes of identification methods for different systems were analyzed in [18,23–26].
The recursive prediction error least-squares method can identify the parameters of Box–Jenkins systems [27], but the
stochastic gradient (SG) identification algorithm has low computational load and slow convergence rates [28]. Recently, Liu,
Wang and Ding presented a least-square-based iterative identification algorithm for Box–Jenkins models [29]. On the basis
of their work in [29], the objective of this paper is to develop new identification algorithms using the iterative techniques and
to present a gradient-based iterative identification algorithm for Box–Jenkins systems to improve the parameter estimation
accuracy.
The paper is organized as follows. Section 2 simply introduces the prediction error stochastic gradient algorithm for
Box–Jenkins models and Section 3 derives a gradient-based iterative identification algorithm for Box–Jenkins systems.
Section 4 gives an illustrative example. Finally, concluding remarks are given in Section 5.
2. The stochastic gradient algorithms
Consider the following Box–Jenkins systems in [29],
y(t) =
B(z)
A(z)
u(t) +
D(z)
C(z)
v(t), (1)
I
This work was supported by the National Natural Science Foundation of China (60973048).
∗
Corresponding author.
E-mail addresses: dqwang64@163.com (D. Wang), ygw_ustb@163.com (G. Yang), rfding@yahoo.cn (R. Ding).
0898-1221/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.camwa.2010.06.001