Work

A Stability-Aware Gap Metric Oriented Framework for Smooth and Stable Blending in Multiple Model Control

Year: 2026
Type: preprint
Abstract: This paper addresses a critical limitation of gap metric-based blending approaches for multiple model control of nonlinear systems. While existing methods compute weighting functions solely based on the open-loop distance between local models, they neglect explicit closed-loop stability criteria during the blending stage. Consequently, even when the local controllers are individually stabilizing, the resulting global controller may compromise overall closed-loop stability. To overcome this shortcoming, a novel stability-aware weighting framework is proposed that explicitly integrates stability conditions into the gap metric-based weights calculation. The proposed method ensures that only stabilizing local controllers contribute to the global multiple model control signal. Furthermore, to prevent hard switching and oscillations that may arise from directly enforcing stability constraints, an auxiliary local model strategy is introduced that enables smooth transitions across operating regions. The effectiveness of the framework is demonstrated through comprehensive simulations on a highly nonlinear continuous stirred tank reactor (CSTR) benchmark over a wide operating range. Results confirm that the proposed approach not only guarantees closed-loop stability but also achieves superior tracking performance compared to conventional gap metric-based methods, while eliminating undesirable oscillations during region transitions through the use of auxiliary local models.
Language: English
Open Access status: green
OpenAlex ID: W7235059756
Cites: 22
Cited by:
Related to: 10