Single-step phase identification and phase locking for coherent beam combination using deep learning

Abstract

Abstract Coherent beam combination offers a solution to the challenges associated with the power handling capacity of individual fibres, however, the combined intensity profile strongly depends on the relative phase of each fibre. Optimal combination necessitates precise control over the phase of each fibre channel, however, determining the required phase compensations is challenging because phase information is typically not available. Additionally, the presence of continuously varying phase noise in fibre laser systems means that a single-step and high-speed correction process is required. In this work, we use a spatial light modulator to demonstrate coherent combination in a seven-beam system. Deep learning is used to identify the relative phase offsets for each beam directly from the combined intensity pattern, allowing real-time correction. Furthermore, we demonstrate that the deep learning agent can calculate the phase corrections needed to achieve user-specified target intensity profiles thus simultaneously achieving both beam combination and beam shaping.