Deep-learning-based coherent fiber laser array system for power scaling and spatial light structuring

In this report, we will introduce our recent advances in developing the deep-learning-based coherent fiber laser array systems for power scaling and spatial light structuring. Our motivation is to construct a deep-learning network for estimating the thermal and environmental induced phase errors, and further compensate the phase errors by the phase control servo with the assistance of the network outputs. Technical progresses in terms of the network optimization, two-stage control scheme, and optical field information acquisition will be covered. Moreover, the prospects and challenges towards the future implementation of intelligent control for CBC systems will be discussed.