标题： 通过置信度优势机动指标进行机动检测 显示英文标题

标题： Maneuver Detection via a Confidence Dominance Maneuver Indicator

摘要： 精确且高效的机动检测对于确保航天器轨迹的安全性和可预测性至关重要。 本文提出了一种基于比较轨道状态估计和观测似然相关置信度的新机动检测方法。 首先，通过为状态估计设置一个置信度并计算观测的最大似然及其置信度，提出了一个置信度主导的机动指示器（CDMI）。 当观测的置信度超过状态估计的置信度时，CDMI会标记一次机动，这表明在无机动假设下观测是不可能的，同时与先验状态估计的置信度保持一致。 为了高效计算观测的最大似然并获得CDMI，开发了一种递归多项式优化方法，利用凸优化和多项式逼近。 此外，还开发了一种集成CDMI方法，以消除手动选择状态置信度的需要。 集成CDMI方法在保持高检测准确性的同时，还能提供机动可能性的指示，从而增强鲁棒性和实际应用性。 所提出的基于CDMI的机动检测方法的性能是针对最优控制距离度量和两种基于混合的方法进行评估的。 仿真结果表明，所提出的集成CDMI方法可以达到高达99.33%的检测准确率，比竞争方法至少高出10%，同时显著降低计算成本。 显示英文摘要

摘要： Accurate and efficient maneuver detection is critical for ensuring the safety and predictability of spacecraft trajectories. This paper presents a novel maneuver detection approach based on comparing the confidence levels associated with the orbital state estimation and the observation likelihood. First, a confidence-dominance maneuver indicator (CDMI) is proposed by setting a confidence level for the state estimation and computing the maximum likelihood of the observation and its confidence level. The CDMI then flag a maneuver when the observation's confidence level exceeds that of the state estimation, indicating that the observation is unlikely under the no-maneuver hypothesis while maintaining consistency with the prior state estimation confidence. To efficiently compute the maximum likelihood of the observation and obtain the CDMI, a recursive polynomial optimization method is developed, taking advantage of convex optimization and polynomial approximation. In addition, an integrated CDMI approach is developed to eliminate the need to manually select the state confidence level. The integrated CDMI approach maintains high detection accuracy while simultaneously providing an indication of maneuver likelihood, thereby enhancing robustness and practical applicability. The performance of the proposed CDMI-based maneuver detection approaches is evaluated against an optimal control distance metric and two mixture-based approaches. The simulation results demonstrate that the proposed integrated CDMI approach can achieve up to 99.33\% detection accuracy, at least 10% higher than the competing methods, while substantially reducing computational costs.