P‐7.2: A Physics‐Informed Multi‐Agent Deep Reinforcement Learning Framework for Decoupling Control in Electrohydrodynamic Inkjet Printing Arrays
SID Symposium Digest of Technical Papers · 2026-08 · 57 (S1) · 1433-1438
DOI: 10.1002/sdtp.70003_382 ↗Kurz gefasst
Fachartikel in der Zeitschrift SID Symposium Digest of Technical Papers, online erschienen am 11.09.2026. Verfasst von Yulin Li und 6 weiteren Autor:innen. Die Arbeit nennt 12 Literaturangaben.
Abstract (Originalsprache)
Electrohydrodynamic inkjet printing is a promising technology for high‐resolution micro‐fabrication, yet achieving uniform deposition in high‐density arrays remains a challenge due to the strong electrostatic coupling between nozzles. Voltage fluctuations in one nozzle inevitably alter the local electric field distribution, disturbing the jet state of adjacent nozzles. Consequently, traditional independent control methods fail to address this non‐linear crosstalk, leading to poor printing consistency.This paper proposes a decoupling control framework based on Physics‐Informed Multi‐Agent Deep Reinforcement Learning (MADRL). The array control process is modeled as a fully cooperative Markov game. To enable agents to perceive environmental coupling, a compact state vector incorporating a heuristic electric field disturbance metric is constructed. A parameter‐sharing Deep Q‐Network (DQN) is employed with a reward function designed to balance individual tracking precision and neighborhood consistency, allowing agents to learn collaborative compensation strategies. The method was validated on a multi‐nozzle printing system. The results indicate that the proposed framework effectively minimizes the droplet volume coefficient of variation (CV) of the full array. Moreover, the system demonstrates strong robustness and high recovery efficiency in the presence of sudden disturbances.
Zitieren
APA
Li, Y., Wang, G., Li, Z., Zhao, Z., Ji, J., Duan, Y., & Chen, J. (2026). P‐7.2: A Physics‐Informed Multi‐Agent Deep Reinforcement Learning Framework for Decoupling Control in Electrohydrodynamic Inkjet Printing Arrays. SID Symposium Digest of Technical Papers, 57(S1), 1433-1438. https://doi.org/10.1002/sdtp.70003_382
Vancouver
Li Y, Wang G, Li Z, Zhao Z, Ji J, Duan Y, et al. P‐7.2: A Physics‐Informed Multi‐Agent Deep Reinforcement Learning Framework for Decoupling Control in Electrohydrodynamic Inkjet Printing Arrays. SID Symposium Digest of Technical Papers. 2026;57(S1):1433-1438. doi:10.1002/sdtp.70003_382
BibTeX
@article{Li2026,
author = {Li, Yulin and Wang, Guozhen and Li, Zhuorun and Zhao, Zheng and Ji, Junwang and Duan, Yongqing and Chen, Jiankui},
title = {P‐7.2: A Physics‐Informed Multi‐Agent Deep Reinforcement Learning Framework for Decoupling Control in Electrohydrodynamic Inkjet Printing Arrays},
journal = {SID Symposium Digest of Technical Papers},
year = {2026},
volume = {57},
number = {S1},
pages = {1433--1438},
doi = {10.1002/sdtp.70003_382}
}
EUVIMED