论文部分内容阅读
由于地球引力和大气阻力等因素造成的模型不确定性,使常规滤波方法用于卫星编队飞行相对运动估计时精度不高。为克服这种影响,提出了一种融合高斯过程回归(Gaussian process regression,GPR)的无模型无迹粒子滤波(model-free unscented particle filter,MF-UPF)方法。对近圆轨道的双星编队问题,新方法通过高斯过程回归对已有的量测数据学习建立相对运动模型,有效地避免了模型不确定性造成的滤波性能下降。仿真对比验证了无模型无迹粒子滤波在编队飞行相对运动估计中的优越性。
Due to the model uncertainty caused by factors such as gravitation and atmospheric drag, the conventional filtering method is not suitable for the relative motion estimation of satellite formation flight. To overcome this effect, a model-free unscented particle filter (MF-UPF) method integrating Gaussian process regression (GPR) is proposed. For the double-star orbit formation near the circular orbit, the new method establishes the relative motion model to the existing measurement data through Gaussian process regression, which effectively avoids the decline of the filtering performance caused by the model uncertainty. Simulation results verify the superiority of modelless trace particle filter in relative motion estimation of formation flying.