基于卡尔曼滤波的AAM飞行管理系统建模
Kalman Filtering Based Flight Management System Modeling for AAM Aircraft
AI总结：
本文提出基于卡尔曼滤波的AAM飞行管理系统建模方法，通过动态调整测量信任提高飞行计划验证的准确性。
摘要：
先进空中交通（AAM）运营需要战略飞行计划服务，该服务需预测空间和时间的不确定性，以安全地将飞行计划验证 against 危险如天气细胞、受限空域和CNS中断区域。当前AAM车辆的不确定性估计方法依赖于保守的线性模型，由于实际世界性能数据有限。本文提出了一种新颖的基于卡尔曼滤波的不确定性传播方法，通过sigmoid混合的测量噪声协方差建模AAM飞行管理系统（FMS）架构。与现有方法固定不确定性阈值不同，我们的方法根据向航路点的进展连续适应滤波器的测量信任，使FMS的纠正行为自然出现。该方法与控制输入成比例扩展，并可调谐以匹配特定飞机特性或航线条件。我们使用真实ADS-B数据验证该方法，数据来自通用航空飞机，分为训练集和验证集。不确定性传播参数在训练集上进行调优，预测到达时间的准确率为76%，与验证数据集相比，证明了该方法在AAM运营战略飞行计划验证中的有效性。
英文摘要：
Advanced Air Mobility (AAM) operations are planned to utilize strategic flight planning services that predict temporal uncertainties to validate flight plans against hazards such as weather cells, restricted airspaces, and CNS disruption areas. This paper presents a Kalman Filter-based uncertainty propagation method that models Flight Management System (FMS) correction behavior through a sigmoid-blended measurement noise covariance. The sigmoid formulation generalizes existing discrete FMS activation thresholds into a continuous, tunable function that smoothly transitions the filter's measurement noise based on progress toward each waypoint. When the measurement noise is high, due to an inverse relationship, the Kalman gain is small and thus uncertainty grows; as the aircraft nears a waypoint, measurement noise decreases as a function of progress, the Kalman gain increases, and state covariance contracts which models the FMS progressively correcting toward the planned trajectory. The approach is computationally efficient (up to two orders of magnitude faster than Monte Carlo methods), scales with control inputs, and is parametrically tunable for different classes of aircraft. The measurement noise covariance is calibrated using real Automatic Dependent Surveillance-Broadcast (ADS-B) data from commercial Instrument Flight Rules (IFR) flights serving as surrogates for future AAM operations, achieving coverage probability conservative relative to theoretical Gaussian predictions at the 1-sigma confidence level on a hold out verification dataset (N = 36). Parameter sensitivity analysis across multiple flight routes demonstrates robust behavior, and comparative evaluation against Monte Carlo and Linear Propagation methods contextualizes the method's computational and accuracy trade-offs.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
