D4Q: Data-Driven Design of Dynamic QuantizerProposal of Method and Experimental Validation

• 成果类型: Article
• 署名作者: Fujimoto, Yusuke; Minami, Yuki
• 署名单位: University of Osaka; University of Hyogo
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2025.3623044
• 发表日期: 2026
• 关键词: feedback stabilization; QUANTIZATION; systems

摘要

This article discusses a data-driven design method for a dynamic quantizer. In particular, we consider the parameter tuning of a noise-shaping filter under the assumption that the input-output data are available but the target plant itself is unknown. We first focus on the input-output relationship of the noise-shaping filter in the optimal dynamic quantizer (ODQ). Based on this relationship, a data-driven parameter tuning method that makes the noise-shaping filter similar to the optimal one is proposed. The quantizer obtained with the proposed method converges to the ODQ under the assumptions: first, the structure of the noise-shaping filter is the same as that of the optimal one, and second, the observed data are informative, i.e., persistently exciting of an order that depends on the noise-shaping filter. The effectiveness of the proposed method is demonstrated through a numerical example and a practical experiment with a dc motor.