Multidirection Gradient Iterative Algorithm: A Unified Framework for Gradient Iterative and Least Squares Algorithms

• 成果类型: Article
• 署名作者: Chen, Jing; Ma, Junxia; Gan, Min; Zhu, Quanmin
• 署名单位: Jiangnan University; Jiangnan University; Qingdao University
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2021.3132262
• 发表日期: 2022
• 页码: 6770-6777
• 关键词: convergence; Eigenvalues and eigenfunctions; computational modeling; mathematical models; Load modeling; Technological innovation; Iterative algorithms; Computational load; convergence rate; gradient iterative (GI) algorithm; Hammerstein system; irregular sampling; multidirection

摘要

In this article, a multidirection-based gradient iterative (GI) algorithm for Hammerstein systems with irregular sampling data is proposed. The algorithm updates the parameter estimates using several orthogonal directions at each iteration. The convergence rate is significantly improved with an increasing number of directions. The convergence property and two simulation examples are provided to demonstrate the effectiveness of the proposed algorithm. In addition, the multidirection-based GI algorithm establishes a relationship between the traditional GI and least squares (LS) algorithms. Thus, our algorithm that combines the LS and GI algorithms constructs an identification framework for a significantly wider class of systems.