Abstract:

One of the largest error sources affecting the Required Navigation Performance （RNP） of the Global Navigation Satellite System （GNSS） is ionospheric delay. This delay is proportional to the Total Electron Content （TEC） of the ionosphere， so the accurate forecasting of TEC directly affects the RNP of GNSS. The Dingo Optimization Algorithm （DOA） proposed in 2021 was used to optimize the Back Propagation （BP） neural network， and a TEC short-term forecasting model was constructed based on the DOA-BP neural network. Using the ionospheric grid dot TEC values provided by the Center for Orbit Determination in Europe （CODE） as the data sets， the forecasting model was trained and tested to achieve high precision short-term forecasting of TEC values at different ionospheric grid points globally and in China. The predicted TEC values were compared with those predicted by the traditional BP neural network model and Sparrow Search Algorithm （SSA） optimized BP neural network model. It was found that compared with the traditional BP model， the DOA-BP model has significantly improved the forecasting accuracy of TEC. Compared with other optimized models （such as SSA-BP model）， the DOA-BP model can also obtain better forecasting accuracy of TEC. The results show that the DOA-BP model can accurately reflect the change characteristics of ionospheric TEC in different time and space around the world， and can be used as a new method for short-term forecasting of ionospheric TEC.

Key words: ionosphere, total electron content （TEC）, dingo optimization algorithm （DOA）, back propagation （BP） neural network, short-term prediction