Improved model order reduction techniques with error bounds
INTERNATIONAL JOURNAL OF SYSTEMS SCIENCE(2024)
摘要
This paper introduces two enhanced model order reduction techniques designed for scenarios involving frequency-weighted and frequency-limited-interval Gramians in the continuous-time domain. The primary objective is to address the instability issue identified in existing approaches in the continuous-time domain, as formulated by Enns for frequency-weighted scenarios and Gawronski & Juang for frequency-limited-interval scenarios. Despite numerous solutions proposed in the literature to mitigate this problem, a persistent challenge remains the high approximation error between the original and reduced-order systems. To overcome this limitation, the proposed improved techniques focus on ensuring stability in reduced-order models while simultaneously minimising the approximation error between the original and reduced systems. Furthermore, these enhanced techniques provide a computationally straightforward, a priori error bound formula. Numerical findings underscore the correctness and efficiency of the proposed techniques in reducing the approximation error while maintaining stability, thereby substantiating their efficacy.
更多查看译文
关键词
Reduced-order model,frequency-limited-interval,frequency-weights,Gramians,error bound
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要