RBFNN-based angles-only orbit determination method with cooperative dual-satellite
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
万方数据
维普
针对空间非合作目标空间态势感知任务中弱可观测无源定轨状态的快速捕获问题,提出了一种基于径向基函数神经网络(RBFNN)的双星协同稀疏无源测角定轨方法.首先,在限制性三体问题的假设下建立了考虑地球非球形 J2 项摄动的轨道动力学模型和赤经赤纬测量模型.然后,构建了基于RBFNN的双星协同仅测角定轨框架,设计了训练数据集生成器、数据预处理方法和RBFNN结构.最后,利用地球静止轨道任务进行了数值仿真验证,并对测角频率等参数的定轨敏感性进行分析.仿真结果表明,在 240 s 内仅进行三次角度观测的条件下,该模型对初始相对距离估计的平均绝对百分比误差约为 0.36%,目标轨道速度的估计误差在米/秒量级,可实现高精度的超短弧段稀疏无源测量定轨.
To solve the rapid capture problem of the weak observability of passive orbit determination state during the non-cooperative target's space situational awareness mission,an angles-only orbit determination method with cooperative dual-satellite based on RBFNN is proposed.Firstly,under the restricted three-body problem hypothesis,an orbital dynamics model considering the Earth's non-spherical J2 perturbation and a model for right ascension and declination measurements are established.Subsequently,a framework for dual-satellite cooperative angles-only orbit determination based on RBFNN is formulated,and the training data set generator,the data preprocessing method,and the structure of RBFNN are designed.Finally,the numerical simulations based on geostationary orbit type mission are conducted.Additionally,sensitivity analysis of orbit determination with respect to parameters such as measurement frequency is conducted.The simulation results show that under the condition of only three angle observations in 240 s,the mean-absolute percentage error in initial relative distance estimation of the model is about 0.36%,and the error of target orbit velocity estimation maintains meter/second-level accuracy,which can realize high-precision passive orbit determination of too-short arc.
space situational awarenessinitial orbit determinationangles-only measurementradial basis function neural networkcooperative dual-satellite
万方数据
维普
