Satellite clock error is a key factor affecting the positioning accuracy of a global navigation satellite system (GNSS). In this paper, we use a gated recurrent unit (GRU) neural network to construct a satellite clock bias forecasting model for the BDS-3 navigation system. In order to further improve the prediction accuracy and stability of the GRU, this paper proposes a satellite clock bias forecasting model, termed ITSSA-GRU, which combines the improved sparrow search algorithm (SSA) and the GRU, avoiding the problems of GRU’s sensitivity to hyperparameters and its tendency to fall into local optimal solutions. The model improves the initialization population phase of the SSA by introducing iterative chaotic mapping and adopts an iterative update strategy based on t-step optimization to enhance the optimization ability of the SSA. Five models, namely, ITSSA-GRU, SSA-GRU, GRU, LSTM, and GM(1,1), are used to forecast the satellite clock bias data in three different types of orbits of the BDS-3 system: MEO, IGSO, and GEO. The experimental results show that, as compared with the other four models, the ITSSA-GRU model has a stronger generalization ability and forecasting effect in the clock bias forecasting of all three types of satellites. Therefore, the ITSSA-GRU model can provide a new means of improving the accuracy of navigation satellite clock bias forecasting to meet the needs of high-precision positioning.

Based on the original deep network architecture, this paper replaces the deep integrated network by integrating shallow FastText, a bidirectional gated recurrent unit (GRU) network and the convolutional neural networks (CNNs). In FastText, word embedding, 2-grams and 3-grams are combined to extract text features. In recurrent neural networks (RNNs), a bidirectional GRU network is used to lessen information loss during the process of transmission. In CNNs, text features are extracted using various convolutional kernel sizes. Additionally, three optimization algorithms are utilized to improve the classification capabilities of each network architecture. The experimental findings using the social network news dataset demonstrate that the integrated model is effective in improving the accuracy of text classification.

The standard products of the International Laser Ranging Service (ILRS) are mainly based on the two laser geodynamics satellites (LAGEOS) due to the sparse observations of the Etalon satellites. With improvements in the ability to track high-altitude satellites, ILRS conducted a 3-month Etalon tracking campaign. In this paper, we study the contribution of more Etalon observations in the new observation scenario to weekly ILRS products, such as station coordinates, Earth rotation parameters (ERPs) and satellite orbit. We compare the ILRS products estimated from LAGEOS-only solutions and LAGEOS+Etalon solutions. In the new observation scenario of 2019, the numbers of observations of Etalon satellites are 1.4 and 1.7 times larger than those in 2018. It is shown that the quality of station coordinates, and the satellite orbit of LAGESOS satellites are only slightly affected by the increase in Etalon observations of the campaign. However, for station 1868, which is dedicated to high-altitude satellites, the root mean square (RMS) values of the residuals in the N, E, and U components are improved by 3.1 cm, 2.1 cm and 2.3 cm, respectively. The internal precision of orbit for Etalon-1/2 satellites in tangle and normal directions are improved by 1.5 cm and 2.9 cm, respectively. Most remarkably, the standard deviations for Xp, Yp and LOD can be improved by 6.9%, 14.3% and 5.1%, respectively, compared with the International Earth Rotation System (IERS)-14-C04 series. With our research, the ILRS could increase efforts on Etalon satellite tracking without affecting the routine observations of LAGEOS satellites.