Research on clock holding technology based on PSO-BP neural network

Abstract

Aiming at the problem that the local crystal oscillator is affected by its own aging factors, which will lead to a decrease in the retention accuracy of the clock synchronization system, firstly, by introducing the particle swarm optimization algorithm, the selection of the initial weight and threshold of the BP neural network is optimized, and the convergence speed is improved. Then use the PSO-BP neural network model to fit and predict the aging data of the two groups of crystal oscillators, establish a related aging model, compare the prediction error of the BP neural network model before and after the optimization of the particle swarm algorithm, and verify the good optimization ability of the particle swarm algorithm. Finally, the model is applied to the clock synchronization system. The frequency accuracy of the system within 24 hours of reference 1PPS signal loss can be maintained at the order of $\pm 8 \times 10^{-11}$, achieving a high-precision clock retention effect.