Citation

Tan, Ai Hui and Foo, Mathias Fui Lin (2025) Kernel design for estimation of resonant systems: A case study on vehicle suspension. Mechanical Systems and Signal Processing, 234. p. 112875. ISSN 08883270

Text Kernel design for estimation of resonant systems_ A case study on vehicle suspension.pdf - Published Version Restricted to Repository staff only Download (1MB)

Abstract

This paper considers the kernel design for impulse response estimation of resonant systems, exemplified through a vehicle suspension system. The identification of resonant systems is highly challenging due to their long impulse responses and dynamics that may be close to the limit of stability. These issues have not been investigated in the existing literature on kernel design. A novel incremental approach for the design of multiple-kernels is proposed, where the kernels are added sequentially in a structured manner, with checks to ensure that they match practical ex pectations. The proposed method possesses the advantages of simplicity, low computational complexity, good numerical conditioning and high feasibility in practical applications. A detailed case study on a vehicle suspension system is presented for the theoretical case and experimental cases with two different grades of road roughness leading to different resonant behavior. It was found that the proposed technique resulted in simple multiple-kernels comprising a combination of Kautz kernels that outperformed the current state-of-the-art diagonal-correlated and single Kautz kernels, recording the highest output fit and the lowest uncertainty under real-world conditions on four separate experimental datasets. Additionally, the incremental approach suc cessfully captured the multiple resonances in the system. Comparison with multiple-kernels of the same structure but optimized using a competing technique showed that the proposed method has an increasing edge as the number of resonances increases. The findings from this work are sig nificant for identifying systems with multiple resonances, such as for structural health monitoring and the development of digital twins.

Item Type:	Article
Uncontrolled Keywords:	Kernels
Subjects:	Q Science > QC Physics > QC 1-75 General
Divisions:	Faculty of Engineering (FOE)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	30 May 2025 01:57
Last Modified:	30 May 2025 01:57
URII:	http://shdl.mmu.edu.my/id/eprint/13884

Downloads

Edit (login required)