Improved Reduced-Order Modeling Using Clustering Method with Dominant Pole Retention

Abstract

This paper provides a quantitative measure criterion for selection of poles from a higher order model. The selection of poles is important because it determines both the transient and steady-state information of the dynamical system. The new indices specify which poles are dominant even when they are not the slowest. On the basis of important poles contribution to the system poles, they are selected to form cluster center. Pade approximation method is used to estimate the coefficients of numerator polynomial. Besides, this technique gives a better approximation in both the transient and the steady-state responses of the large-scale system. The effectiveness of the proposed method is demonstrated through numerical test examples and compared with other well-known published methods.

KEYWORDS:

• Order reduction
• Clustering
• Dominant pole
• Pade approximation
• Integral square error

Additional information

Notes on contributors

Sharad Kumar Tiwari

Sharad Kumar Tiwari received the BTech degree in electronics & instrumentation engineering in 2009 from U.P.T.U. and the ME degree in electronic instrumentation & control engineering in 2011 from Thapar University, India. He is currently pursuing PhD degree in the Department of Electrical & Instrumentation Engineering, Thapar University, India. His research interests include modeling and control of linear systems, model order reduction, artificial intelligent systems and applications of control theory. Corresponding author. Email: [email protected]

Gagandeep Kaur

Gagandeep Kaur received BTech degree in instrumentation engineering from NIT, Jalandhar 1997 and ME degree in electronic instrumentation & control engineering from Thapar University 2001 and PhD degree from Thapar University, in 2012. Since 1997, she is working in Department of Electrical and Instrumentation Engineering, Thapar University, India. Her research interests are intelligent systems design, soft computing, artificial intelligent systems, modeling and identification, model order reduction, and renewable energy. Email: [email protected]