发表机构
University of Dhaka; La Trobe University（达卡大学; 拉筹伯大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究针对CubeSat在轨故障无法及时修复导致寿命不足的问题，提出PHOENIX系统，利用微调SLM与多智能体AI实现自主故障修复，基于ESA基准验证了初步效果。
AI 中文摘要
大多数CubeSat（约鞋盒大小的小型低成本卫星）的实际寿命达不到设计预期：一项针对178次任务的研究发现，两年后仍正常运行的卫星仅占48%-65%，而其设计寿命为2-5年。更核心的问题在于，近地轨道（LEO）上的CubeSat在每次96分钟的轨道周期中，约有85分钟无法与地面建立物理连接，因此在此期间发生的故障会被忽略，直到下一次通信时段才会被发现，此时可能已无法修复。我们提出PHOENIX（Predictive Health On-orbit Edge Neural Intelligence eXtension，即轨道边缘神经智能预测健康扩展），为卫星赋予自主故障推理能力。一款经微调的小型语言模型（SLM）体积足够小，可在嵌入式硬件上运行，部署在CubeSat上，运行于经飞行验证的Aethero NxN-ECM计算机，持续监测所有传感器读数，并利用存储过往修复记录的内存系统解决重复性故障，避免重复运行相同推理过程。每轨道周期内，卫星会向地面发送一份简短的结构化健康报告，而非原始数据转储；地面的六个专用AI智能体将在5-10分钟的通信窗口内读取该报告并生成经过验证的卫星指令。由于真实故障示例仅占数据集的0.57%-1.80%，我们采用生成式扩散模型（DDPM）创建合成训练数据。我们在ESA异常检测基准（涵盖14年数据、76个通道、118个标注故障）上报告了初步结果。
英文摘要
Most CubeSats, small and low-cost satellites roughly the size of a shoebox, do not survive as long as they were designed to: a study of 178 missions found that only 48-65% remain operational after two years, against a designed lifetime of 2-5 years. The deeper issue is that a CubeSat in low Earth orbit (LEO) is physically unreachable from the ground for roughly 85 minutes out of every 96-minute orbit, so faults that start during that window go unnoticed until the next contact pass, by which point recovery may no longer be possible. We propose PHOENIX (Predictive Health On-orbit Edge Neural Intelligence eXtension) to give the satellite its own fault reasoning capability. A fine-tuned Small Language Model (SLM) compact enough to run on embedded hardware is deployed onboard the CubeSat, running on the flight-proven Aethero NxN-ECM computer, monitoring all sensor readings continuously, and resolving recurring faults using a memory system that stores past repairs so the same inference does not need to run twice. Once per orbit it sends a short structured health report to the ground instead of a raw data dump; six specialized AI agents on the ground read that report and generate validated satellite commands within the 5-10 minute contact window. A generative diffusion model (DDPM) creates synthetic training data because real fault examples make up only 0.57-1.80% of the dataset. We report preliminary results on the ESA Anomaly Detection Benchmark (14 years, 76 channels, 118 labeled faults).
Comments6 pages, 2 figures. Accepted at IEEE IRAI 2026 (International Conference on Responsible Artificial Intelligence)
