50 / 2023-03-30 17:26:15

Research on fast satellite selection algorithm based on improved NSGA III

Satellite selection algorithms; NSGA III; RDOP; multi-system combinations; multi-objective optimization algorithms

The total number of satellites in the multi-constellation GNSS has now reached more than 140, and the increase in the number of visible satellites has improved the positioning accuracy of the receiver. However, this has increased the computational processing burden on the receiver, which affects the real-time positioning performance. In order to improve the real-time positioning of the receiver and reduce the processing burden of the receiver for multiple constellation combinations, it is necessary to select the constellation combinations with good geometric layout from a large number of visible satellites for positioning settlement. This paper proposes an improved non-dominated sorting genetics algorithm (Ⅲ) to improve the speed of satellite selection while ensuring the positioning accuracy of multi-constellation GNSS. The algorithm uses an elite retention mechanism, fast non-dominated sorting and a diversity preservation strategy to calculate the distance to the reference point. Firstly, the satellites visible at the current epoch are continuously coded and grouped according to the number of selected satellites. The population is generated using the sequential nature of the selection problem, and the constellations to be selected are clustered and analysed into subsets. Individual selection is performed by means of uniformly distributed reference points as a benchmark, from which the optimal subset set is selected. Finally, the optimal solution that meets the user's requirements is obtained by the preference solution function. The results show that for three different cut-off altitude angles of 5°, 15° and 30° for the combined BDS/GPS/Galileo system, the selection algorithm is able to stabilise the number of selected satellites at around 10, significantly reducing the number of satellites selected. Compared to the minimum geometric accuracy factor traversal method, the difference between the RDOP values before and after selection is within 0.2, which maintains a high accuracy and improves the selection time by about 75%. The algorithm is suitable for the case of multiple GNSS systems with different numbers of selected satellites for combined navigation positioning, and can be used for continuous positioning in a more efficient way.