A Novel High-Precision Method Based on Adaptive Unscented Particle Filter for BeiDou/GPS Disciplined Crystal Oscillator

There is a random error in the pulse per second (PPS) signal output by BeiDou/GPS receiver. Meanwhile, the error of the frequency difference signal exists because of the frequency nonlinear drift of crystal oscillator caused by the module aging. The mathematical model of the frequency difference between the PPS signal and crystal oscillator signal is established, and the adaptive unscented particle filter(A-UPF) algorithm is proposed to process the noise contained in the frequency difference signal. In the importance sampling stage, the unscented Kalman filter is used to generate the optimal distribution function, which can reduce the estimation error and increase the robustness of the system. In the measurement update phase, the attenuation factor is introduced to reduce the impact of the uncertainty of noise statistical characteristics on the filtering result and overcome the phenomenon of particle degradation in UPF. When the statistical characteristics of the system noise are determined, the realtime performance of the adaptive UPF is also improved. Through simulation verification, the accuracy of output frequency can be improved by applying the A-UPF algorithm to BeiDou/GPS disciplining crystal oscillator system.