Citation

Hii, S. J. and Tan, Ai Hui and Cham, Chin Leei (2025) Finite Impulse Response Identification Using Kernel-Based Regularization for a Cooling System. In: 5th IEEE International Conference on System Engineering and Technology, ICSET 2025, 4 October 2025, Kuala Lumpur.

Text 89.pdf - Published Version Restricted to Repository staff only Download (1MB)

Abstract

This paper considers the identification of finite impulse response (FIR) models for the flow and Peltier modules in a cooling system benchmark using kernel-based regularization. The kernels investigated include the oracle kernel, the tuned-correlated kernel and the diagonal-correlated kernel. The oracle kernel is constructed from an FIR model obtained by truncating the responses of an autoregressive with exogenous input model or an output error model. Results show that the diagonal-correlated kernel is overall superior to the other designs despite the flow and Peltier modules having significantly different dynamics. For the flow module, the diagonal-correlated kernel achieved a mean square error reduction of 7.9% from that of the classical prediction error model. The corresponding reduction for the Peltier module was 25%. This improvement is due to the diagonal-correlated kernel having high flexibility, as it has three hyperparameters that enable it to capture challenging dynamics in the cooling system. The findings from this work can be applied to improve the identification of practical systems, especially thermodynamic systems such as battery packs.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Estimation, impulse responses, kernels
Subjects:	Q Science > QC Physics > QC 1-75 General
Divisions:	Faculty of Artificial Intelligence & Engineering (FAIE)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	18 Mar 2026 08:06
Last Modified:	06 Apr 2026 04:32
URII:	http://shdl.mmu.edu.my/id/eprint/15575

Downloads

Edit (login required)