A multi-model modeling approach to nonlinear systems based on lazy learning

被引：0

作者：

Pan Tianhong ^{[1
]}

Li Shaoyuan ^{[1
]}

Wang Xin ^{[1
]}

机构：

[1] Shanghai Jiao Tong Univ, Inst Automat, Shanghai 200030, Peoples R China

来源：

Proceedings of the 24th Chinese Control Conference, Vols 1 and 2 | 2005年

关键词：

lazy learning; nonlinear system; multi-model; recursive least squares; K-Vector Neighbors;

D O I：

暂无

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

Based on local modeling algorithm and from input-output data set, an information select rule(K-VN) was presented for unknown structure nonlinear systems. According to this rule, the learning set was constructed. Using RLS Algorithm and PRESS, the local model of system was built. With the change of working points, multiple local models are built, which can realize the exact modeling for the global system. Simulation results show good performances of the method's simple, effective and reliable estimation.

引用

页码：268 / 273

页数：6

共 6 条

[1]

Atkeson CG, 1997, ARTIF INTELL REV, V11, P11, DOI 10.1023/A:1006559212014

[2] The local paradigm for modeling and control: from neuro-fuzzy to lazy learning [J].

Bontempi, G ;

Bersini, H ;

Birattari, M .

FUZZY SETS AND SYSTEMS, 2001, 121 (01) :59-72

[3]

Bontempi G, 1999, INT J CONTROL, V72, P643, DOI 10.1080/002071799220830

[4]

Murry- Smith R., 1997, MULTIPLE MODEL APPRO

[5] Intelligent selection of instances for prediction functions in lazy learning algorithms [J].

Zhang, JP ;

Yim, YS ;

Yang, JM .

ARTIFICIAL INTELLIGENCE REVIEW, 1997, 11 (1-5) :175-191

[6]

ZHENG Q, 2001, ASIAN J CONTROL, V3, P35

← 1 →