Structured, Reduced-Order H2-Conic Control

• 成果类型: Article
• 署名作者: LoCicero, Ethan J.; Bridgeman, Leila J.
• 署名单位: Duke University
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2023.3303489
• 发表日期: 2024
• 页码: 1356-1363
• 关键词: Control system synthesis; decentralized control; Distributed control; Linear matrix inequalities (LMIs); large-scale systems; Nonlinear systems; Semidefinite programming

摘要

Practical controllers for large-scale systems must be low-order and sparsely communicating, as well as high-performing and robust to uncertainty. H-2-conic design addresses the latter requirements where passivity and H(infinity )methods are inapplicable, but it has yet to address the former. Here, the fixed-order H-2-conic design problem is posed as a series of convergent approximations. The resulting synthesis algorithm uses all controller parameters as explicit design variables, and is guaranteed to find a feasible controller under mild assumptions. This improves performance over previous H-2-conic designs while permitting controller dimension and communication structure to be arbitrarily chosen, which allows reduced-order and sparse distributed designs. The relationship between variable structure and computational complexity is explored, and the main algorithm is applied to numerical examples.