Fig. 1. Device and result diagrams of coherent combining system^[39]. (a) Device diagram of two-channel combining system; (b) change of combined output power in time domain; (c) power spectral density of phase noise

Fig. 2. Model and results of tiled aperture coherent combining system^[40]. (a) Schematic of tiled aperture coherent combination of multi-core fiber laser by deep reinforcement learning; (b) phase errors under different neural networks

Fig. 3. Experimental verification and results of array phase-locked system^[41]. (a) Flow diagram of array phase-locked system; (b) phase quality versus correction step in 100-beam co-phase experiment

Fig. 4. Phase locking experiment and results of coherent beam array ^[42]. (a) Schematic of system for phase-locking coherent beam arrays with neural networks; (b) time domain response of normalized light intensity before and after closing loops; (c) probability distribution of normalized photodetector response values before and after closing loops; (d) corresponding power spectral density before and after closing loops

Fig. 5. Experimental structure and results^[43]. (a) Experimental setup; (b) comparison of SPGD algorithm and Q learning algorithm

Fig. 6. Device diagram and results^[47]. (a) Device diagram; (b) one-dimensional intensity distribution diagram and (c) variation trend diagram of power in bucket under different conditions

Fig. 7. Structural diagram and results of neural networks^[49]. (a) Structural diagram of neural network; (b) normalized combining efficiencies of SPGD and neural network versus number of steps

Fig. 8. Implementation and result diagrams of neural network^[50]. (a) General block diagram of DDRM-based coherent composite stabilizer; (b) combining efficiency versus number of algorithm steps at drift rate of 5°; (c) combining efficiency versus number of algorithm steps at drift rate of 10°

Fig. 9. Experiment and results of tiled aperture coherent combining ^[51]. (a) Diagram of tiled aperture coherent combining system; (b) 7-channel PIB scatter graph obtained under open loop; (c) 19-channel PIB scatter graph obtained under open loop; (d) 7-channel PIB scatter graph obtained by using neural network trained with MSE; (e) 19-channel PIB scatter graph obtained by using neural network trained with MSE; (f) 7-channel PIB scatter graph obtained by using neural network trained with MSE-NPCD; (g) 19-channel PIB scatter graph obtained by using neural network trained with MSE-NPCD

Fig. 10. Experimental structure diagram and results^[54]. (a) Diagram of experiment for stabilizing laser beam combination;(b)-(e) comparison of combining effect between NN and SPGD

Fig. 11. Experimental diagram and results^[63]. (a) Schematic of generating OAM beam with deep learning-assisted two-step phase control method; (b) convergence curves of evaluation functions for generated OAM beams with different topological charges; OAM purity obtained after one step control when N_TC is (c) -1, (d) 1, (e) 2 ;OAM purity obtained after two step control when N_TC is (f) -1, (g) 1, (h) 2

Fig. 12. Experimental diagram and comparison of results^[64]. (a) Schematic of laser array system that adjusts OAM beams by deep learning-based phase control; (b) OAM topology charge after one-step control in 100 simulated cases; (c) OAM topology charge after two-step control in 100 simulated cases; (d) average OAM topology charge after one-step control; (e) average OAM topology charge after two-step control