Citation

Xu, Zhihan and Feng, Liang and Ting, Choo Yee and Gerontitis, Dimitrios K. and Shi, Yang and Wang, Jiyun (2026) A Unified Stencil-Selection Framework for Perturbation-Resilient Discrete Zeroing Dynamics in Time-Varying Quadratic Programming. In: 18th International Conference on Advanced Computational Intelligence, ICACI 2026, 13 August 2026 - 16 August 2026, Tianshui.

Text 4.pdf - Published Version Restricted to Repository staff only Download (1MB)

Abstract

Solvers for time-varying quadratic programming (TVQP) in physical systems face measurement noise and exogenous disturbances that degrade steady-state accuracy. For zeroing neural dynamics (ZND), which converts optimality tracking into recursive online updates, sampled-data performance depends jointly on discretization accuracy and perturbation rejection. This work presents a unified stencil-selection framework that subsumes existing ZND discretizations as instances of a single design parameter, the backward stencil width. Three representative members are derived via Taylor matching and compared under noise-free, constant-noise, and linearly time-varying noise conditions. Experiments confirm that increasing the stencil width basically reduces the steady-state residual monotonically, with the widest-stencil member performing best or comparably across all scenarios.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Zeroing neural dynamics , time-varying quadratic programming
Subjects:	Q Science > QA Mathematics > QA71-90 Instruments and machines > QA75.5-76.95 Electronic computers. Computer science
Divisions:	Faculty of Computing and Informatics (FCI)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	01 Oct 2026 03:10
Last Modified:	01 Oct 2026 03:10
URII:	http://shdl.mmu.edu.my/id/eprint/16763

Downloads

Edit (login required)