Satellite clock error time series prediction method and system based on deep reinforcement learning

Abstract

The present application relates to the technical field of satellite navigation and time synchronization, and particularly relates to a satellite clock error time sequence prediction method and system based on deep reinforcement learning. The present application adopts a noise-gated convolution module to extract features, which includes a data path, a gating path and a noise perception path; the noise perception path is used to extract local frequency jitter of a historical clock error sequence and perform weighted statistics to obtain background noise intensity; the gating path generates adaptive gating coefficients in combination with the background noise intensity; the data path is used to perform dilated convolution processing on the historical clock error sequence to obtain convolution features, and then multiply the convolution features with the adaptive gating coefficients to obtain effective features as the output of the noise-gated convolution module. The present application can perceive the background noise intensity of satellite clock error data in real time; by dynamically adjusting the gating threshold, it can filter false features in a strong noise environment and retain small clock error features in a low noise environment, thereby improving the noise resistance in a non-stationary environment.

Images