Particle Swarm Optimization LSTM Neural Network for Navigation Satellite Clock Bias Forecasting

To tackle the issue of low forecast accuracy in navigation satellite clock bias forecast models, this paper proposes an LSTM neural network clock bias forecast model that utilizes the particle swarm optimization (PSO) algorithm. The model employs the PSO algorithm to seek optimal values for the structural parameters of the LSTM neural network, which reduces the likelihood of local extremes in the LSTM model. The model is then tested using precision clock bias data from the Wuhan International GNSS Service (IGS) center. The forecast results are compared and analyzed against those of the traditional gray GM (1, 1) model, the Autoregressive Integrated Moving Average (ARIMA) model, the Quadratic Polynomial (QP) model, and the LSTM neural network model. The results demonstrate that the accuracy of the LSTM neural network model optimized by the particle swarm algorithm is superior to the other models mentioned in this paper for clock bias forecasting. The satellite clock bias forecasting method based on the particle swarm algorithm optimized LSTM neural network provides a new approach to enhance the accuracy of satellite clock bias forecasting.