【摘要】 低轨互联网星座发展迅速，在星座的设计优化与运行维护阶段均需进行快速、综合的效能评估。针对当前星座效能评估领域存在的评估维度片面、评估权重固化和数据标注成本高等问题，提出一种基于半监督学习的低轨互联网星座效能评估算法。首先，构建一个综合考虑星座覆盖能力、通信质量、服务可靠性等多个层面的效能评估体系，解决单一指标评估分析片面性的问题。然后，基于半监督学习框架，采用渐进式权重策略优化无标签数据的贡献，并引入多尺度注意力网络，通过跨尺度特征融合与注意力机制，提高模型对重要特征的敏感度，实现对星座性能的精准评估。最后，开展低轨互联网星座效能评估仿真验证，仿真结果表明该算法对星座效能评估的最大偏差为5.86%，与层次分析法和监督学习相比，该算法优选出的星座方案性能更佳，同时可实现对星座的准确监测和预警，从而验证了所提效能评估算法的有效性。更多还原

【Abstract】 The development of low-Earth orbit(LEO) internet constellations progresses rapidly, and comprehensive effectiveness assessments must be conducted during both the design optimization and operational maintenance phases. To address the current challenges in constellation effectiveness assessment—such as limited evaluation dimensions, rigid evaluation weights, and high data labeling costs—an effectiveness assessment algorithm for LEO internet constellations based on semi-supervised learning is proposed. First, a comprehensive effectiveness assessment framework is established, incorporating multiple dimensions including constellation coverage capability, communication quality, and service reliability, to overcome the limitations of single-indicator evaluations. Then, based on a semi-supervised learning framework, a progressive weighting strategy is employed to optimize the contribution of unlabeled data, and a multi-scale attention network is introduced. Through cross-scale feature fusion and attention mechanisms, the model’s sensitivity to important features is enhanced, enabling precise assessment of constellation effectiveness. Finally, simulation verification is performed for the proposed LEO internet constellation effectiveness assessment algorithm. Simulation results show that the maximum deviation of the algorithm in assessing constellation effectiveness is 5. 86%. Compared with the analytic hierarchy process and supervised learning methods, the constellation schemes selected by this algorithm exhibit superior performance. Additionally, the algorithm enables accurate monitoring and early warning of constellation effectiveness, thus validating its practical utility.更多还原