Dynamic intelligent frequency locking system based on dual-template kernelized correlation filter

Abstract

Cold atomic gravimeter demonstrates outstanding performance in marine dynamic measurement due to its high accuracy and long-term stability, while the system is sensitive to environmental factors and its laser frequency stabilization system relies on the peaks of individual typical Lamb dip, which are affected by these factors, leading to significant fluctuations in the differential signal and then introduces excessive dead time. This paper reported an automatic laser locking scheme utilizing the artificial intelligence algorithm. A dual-template kernelized correlation filter (DT-KCF) algorithm has been developed to identify the preset peak point of the saturation absorption spectroscopy (SAS) signal by compact optical pathway. Distributed Feedback Laser (DFB) at 780 nm is employed to experimentally investigate the intelligent algorithm through an Rb cell and the D2 saturation absorption lines of ⁸⁷Rb. Modulation-demodulation and incremental PI closed-loop feedback were implemented to fine-tune frequency, and enhanced frequency stabilization accuracy. The frequency locking status was monitored in real-time, and automatically started the re-finding and modem and demodulation links once the lock is unlocked. It can be concluded that the system could automatically relock within about 4.2 s in no human intervention, and the long-term frequency stabilization accuracy is 10^–12.