BDS-2 Broadcast Ephemeris Orbit Error Compensation Based on ABC-BP Neural Network

With the completion of BeiDou Navigation Satellite System (BDS) constellation deployment, the research on improving the service performance of BDS has become the focus. Broadcast ephemeris is a prerequisite for normal navigation in real time navigation and positioning. The real-time performance of broadcast ephemeris results in bad orbit accuracy, and its error variation cannot be accurately described by mathematical model, which has a unignorable impact in positioning and navigation. Based on this research background, we take Multi-GNSS Experiment (MGEX) precision ephemeris as reference and use Back Propagation (BP) Neural Network (NN) and Artificial Bee Colony Algorithm (ABC) optimized BP NN to model the orbit error of broadcast ephemeris to improve the its orbit accuracy. The experimental results show that the two NN models have good modeling effect on broadcast ephemeris orbit error, and the orbit accuracy of broadcast ephemeris of different satellites is improved by 20%–80%. ABC-BP NN has higher accuracy than BP NN, and the results of 3 and 5 days of different orbit types show that it has better ability to control error accumulation, and has better error compensation results for broadcast ephemeris in short and medium term.