基于微波辐射测量的电波折射修正技术

针对电波折射修正精度直接影响无线电系统的探测和定位精度这一问题,提出了利用微波辐射计反演大气折射率剖面进行电波折射修正的解决方法,并引入RBF（Radial Basis Function,径向基）神经网络算法反演大气折射率.在青岛市气象局架设MP-3000A型多通道微波辐射计,开展了长达1个月的与探空数据的联合观测比对实验,对输出的大气折射率剖面进行了详细的分析.实验结果表明：RBF神经网络算法与MP-3000A自带的神经网络算法相比,反演大气折射率剖面的精度提高了30％以上,同时,电波折射修正的精度也得以提高.因此,利用微波辐射计反演大气折射率剖面进行电波折射修正方法可行.     

Autonomous satellite navigation using starlight refraction angle measurements

An on-board autonomous navigation capability is required to reduce the operation costs and enhance the navigation performance of future satellites. Autonomous navigation by stellar refraction is a type of autonomous celestial navigation method that uses high-accuracy star sensors instead of Earth sensors to provide information regarding Earth’s horizon. In previous studies, the refraction apparent height has typically been used for such navigation. However, the apparent height cannot be measured directly by a star sensor and can only be calculated by the refraction angle and an atmospheric refraction model. Therefore, additional errors are introduced by the uncertainty and nonlinearity of atmospheric refraction models, which result in reduced navigation accuracy and reliability. A new navigation method based on the direct measurement of the refraction angle is proposed to solve this problem. Techniques for the determination of the refraction angle are introduced, and a measurement model for the refraction angle is established. The method is tested and validated by simulations. When the starlight refraction height ranges from 20 to 50 km, a positioning accuracy of better than 100 m can be achieved for a low-Earth-orbit (LEO) satellite using the refraction angle, while the positioning accuracy of the traditional method using the apparent height is worse than 500 m under the same conditions. Furthermore, an analysis of the factors that affect navigation accuracy, including the measurement accuracy of the refraction angle, the number of visible refracted stars per orbit and the installation azimuth of star sensor, is presented. This method is highly recommended for small satellites in particular, as no additional hardware besides two star sensors is required.