Model matematik adalah penting bagi mendapatkan sambutan dinamik sesuatu sistem. Maklumat yang diperoleh boleh digunakan untuk mengkaji dan menganalisis sistem tersebut. Pemodelan yang dilakukan berdasarkan data masukan dan keluaran bagi sesuatu sistem dikenali sebagai pengenalpastian sistem. Salah satu isu di dalam pengenalpastian sistem ialah anggaran parameter dan pemilihan struktur model di mana berbagai kaedah telah digunakan termasuk kuasa dua terkecil ortogon. Algoritma kuasa dua terkecil ortagon adalah satu algoritma yang boleh menentukan struktur sesebuah model dengan memberikan sebutan signifikan yang menyumbang kepada model tersebut dan juga dapat memberikan nilai kepada parameter yang dianggarkan. Penerbitan algoritma ini disampaikan dan aplikasinya kepada pemodelan sistem suspensi kereta juga dibincangkan untuk membuktikan keberkesanan algoritma ini.
Kata kunci: Algoritma kuasa dua terkecil ortogon; model sistem dinamik; pengenalpasti sistem; model sistem suspensi.
A mathematical model is important to find a dynamic response of a system. The information obtained from the model could be used for investigation and analysis. Modelling based on input and output data is known as system identification. One of the issues in system identification is the parameter estimation and model structure selection where various methods have been studied including the orthogonal least square algorithm. Orthogonal least square estimation is an algorithm which can determine the structure of a model by identifying the significant terms contributed to the model and also capable of providing the final values of parameter estimates. The derivation of the algorithm is presented and its application to the modelling of a car suspension system is included the demonstrate the effectiveness of the algorithm.
Keywords: Orthogonal least square algorithm; dynamic system modelling; system identification;suspension system model.
[1]
Lennart Ljung,et al.
System Identification: Theory for the User
,
1987
.
[2]
Hussain N. Al-Duwaish,et al.
Nonlinear System Identification Using Neural Networks
,
2003
.
[3]
Ken Sharman,et al.
Genetic algorithms for maximum likelihood parameter estimation
,
1989,
International Conference on Acoustics, Speech, and Signal Processing,.
[4]
Shang-Liang Chen,et al.
Orthogonal least squares learning algorithm for radial basis function networks
,
1991,
IEEE Trans. Neural Networks.
[5]
S. Billings,et al.
Correlation based model validity tests for non-linear models
,
1986
.
[6]
P. N. Paraskevopoulos,et al.
Modern Control Engineering
,
2001
.
[7]
Sheng Chen,et al.
Orthogonal least squares methods and their application to non-linear system identification
,
1989
.
[8]
Andrew G. Alleyne,et al.
Nonlinear adaptive control of active suspensions
,
1995,
IEEE Trans. Control. Syst. Technol..
[9]
Stephen A. Billings,et al.
A comparison of the backpropagation and recursive prediction error algorithms for training neural networks
,
1991
.
[10]
S. Billings,et al.
Orthogonal parameter estimation algorithm for non-linear stochastic systems
,
1988
.
[11]
Huei Peng,et al.
A novel active suspension design technique-simulation and experimental results
,
1997,
Proceedings of the 1997 American Control Conference (Cat. No.97CH36041).