학술논문
CMA-ES/SPGD 이중 알고리즘을 통한 결맞음 빔 결합 시스템 위상제어 및 동작성능에 대한 전산모사 분석
Hybrid CMA-ES/SPGD Algorithm for Phase Control of a Coherent Beam Combining System and its Performance Analysis by Numerical Simulations
Hybrid CMA-ES/SPGD Algorithm for Phase Control of a Coherent Beam Combining System and its Performance Analysis by Numerical Simulations
Document Type
Article
Source
한국광학회지, 34(1), pp.1-12 Feb, 2023
Subject
Language
한국어
ISSN
2287-321X
1225-6285
1225-6285
Abstract
본 연구에서는 다채널 결맞음 빔결합 시스템을 위한 위상제어 방식으로 covariant matrix adaption evolution strategy (CMA-ES) 알고 리즘 및 stochastic parallel gradient descent (SPGD) 알고리즘을 결합한 이중 위상제어 알고리즘을 제안하고 그 동작 특성을 전산모사를 통해 분석한다. 제안하는 CMA-ES/SPGD 이중 위상제어 알고리즘은 결합된 최종 출력광 세기가 미리 설정된 특정값에 도달하기 전까지는 그위상제어 최적화를 CMA-ES 알고리즘을 통해 진행하고, 그 이후에는 SPGD 알고리즘으로 전환하여 진행하는 순차적 이중 구조를 취한다. 이를 이상적인 7채널과 19채널 광섬유 결합기 기반 결맞음 빔결함 시스템에 적용하였을 때, 위상제어 최적화 평균 수렴시간이 기존의 SPGD 알고리즘만 단독적용한 경우에 비해 약 10% 단축됨을 확인하였다. 뿐만 아니라, 동일한 결맞음 빔결함 시스템에서 실제 환경과 유사하게 각 채널광에 위상잡음을 부가적으로 인가한 경우, 본 연구에서 제안하는 이중 위상제어 알고리즘을 적용할 경우 주어진 조건에서 그 평균 수렴시간이 기존의 SPGD 알고리즘만 단독적용한 경우에 비해 7채널 시스템의 경우 약 17%, 19채널 시스템의 경우 약 16–27% 정도 단축됨을 확인하 였다. 본 연구에서 제안한 CMA-ES/SPGD 이중 위상제어 알고리즘은 향후 실제 대기 환경과 같이 위상잡음 효과를 무시할 수 없는 조건에서 결맞음 빔결합을 구현시 매우 유용하게 활용될 수 있을 것으로 기대된다.
In this study, we propose a hybrid phase-control algorithm for multi-channel coherent beam combining (CBC) system by combining the covariant matrix adaption evolution strategy (CMA-ES) and stochastic parallel gradient descent (SPGD) algorithms and analyze its operational performance. The proposed hybrid CMA-ES/SPGD algorithm is a sequential process which initially runs the CMA-ES algorithm until the combined final outputintensity reaches a preset interim value, and then switches to running the SPGD algorithm to the end of the whole process. For ideal 7-channel and19-channel all-fiber-based CBC systems, we have found that the mean convergence time can be reduced by about 10% in comparison with the casewhen the SPGD algorithm is implemented alone. Furthermore, we analyzed a more realistic situation in which some additional phase noise wasintroduced in the same CBC system. As a result, it is shown that the proposed algorithm reduces the mean convergence time by about 17% for a 7-channel CBC system and 16–27% for a 19-channel system compared to the existing SPGD alone algorithm. We expect that for implementing aCBC system in a real outdoor environment where phase noise cannot be ignored, the hybrid CMA-ES/SPGD algorithm proposed in this study will be exploited very usefully.
In this study, we propose a hybrid phase-control algorithm for multi-channel coherent beam combining (CBC) system by combining the covariant matrix adaption evolution strategy (CMA-ES) and stochastic parallel gradient descent (SPGD) algorithms and analyze its operational performance. The proposed hybrid CMA-ES/SPGD algorithm is a sequential process which initially runs the CMA-ES algorithm until the combined final outputintensity reaches a preset interim value, and then switches to running the SPGD algorithm to the end of the whole process. For ideal 7-channel and19-channel all-fiber-based CBC systems, we have found that the mean convergence time can be reduced by about 10% in comparison with the casewhen the SPGD algorithm is implemented alone. Furthermore, we analyzed a more realistic situation in which some additional phase noise wasintroduced in the same CBC system. As a result, it is shown that the proposed algorithm reduces the mean convergence time by about 17% for a 7-channel CBC system and 16–27% for a 19-channel system compared to the existing SPGD alone algorithm. We expect that for implementing aCBC system in a real outdoor environment where phase noise cannot be ignored, the hybrid CMA-ES/SPGD algorithm proposed in this study will be exploited very usefully.