面向海上無(wú)人系統(tǒng)的邊緣模型協(xié)同與數(shù)據(jù)壓縮算法

打開(kāi)文本圖片集

中圖分類(lèi)號(hào)：TP301 文獻(xiàn)標(biāo)志碼：A DOI：10.12305/j.issn.1001-506X.2025.05.34

Abstract：In sea environment，unmanned system relies on edge artificial inteligence（AI）model collaboration to implement data collection and edge processing tasks.However，facing with problems such as poor communication links，limited communication bandwidth，and sensitive to interferences，this paper firstly proposes a model collaborative training method，federated mutual distillation，to reduce thebandwidth requirements formodel training data，from the perspective of AI model collaboration among unmanned aerial vehicle.Secondly，from the perspectiveof data de-redundancy transmision，a data differential dynamic compression method is proposed to reduce the frequency of data transmisson.Simulation results show that the performance of the model，trained with the federated mutual distilation training method，is beter than thatof the benchmarks，and acostof communication bandwidth is reduced compared to the centralized training models. The proposed data diferential dynamic compression method can greatly reduce the sending length and frequency of communication messages，and adapt to the bandwidth bottleneck in the weak communication connection environment.

Keywords：weak communication link；edge model collaboration；federated mutual distillation；data compression

0 引言

近年來(lái)，隨著海洋技術(shù)的發(fā)展，對(duì)于海上船舶信息監(jiān)測(cè)的需求也在不斷提高，然而在海上架設(shè)監(jiān)測(cè)設(shè)備難度過(guò)大。（剩余19078字）