Integration of Singular Spectrum Analysis and Adaptive Neuro-FuzzyInference System for BDS-3 Clock Bias Prediction.

In view of the nonlinear, non-stationary and high-noise characteristics of satellite clock bias (SCB) data, we propose a new clock bias prediction model by integrating adaptive neuro-fuzzy inference system (ANFIS) with singular spectrum analysis (SSA). The developed model is compared with traditional approaches, including grey model (GM (1,1)), quadratic polynomial model (QP), long short-term memory neural network model (LSTM) and ANFIS. The results demonstrate that the SSA-ANFIS model significantly enhances the accuracy and stability of clock bias predictions. In comparison to the ANFIS and LSTM models, the predictive accuracy improves by 45.6% and 32.0%, respectively.