Coherent　beam　combination　(CBC)　is　an　effective　method　to　break　the　limiting　power　of　a　single　fiber　laser.　The　Q-learning　algorithm　is　one　of　the　reinforcement　learning　algorithms.　We　use　the　Q-learning　algorithm　to　do　phase　compensation　in　the　field　of　CBC.　The　performance　difference　between　the　Q-learning　algorithm　and　the　stochastic　parallel　gradient　descent　optimization　algorithm　(SPGD)　is　analyzed　by　simulating　time-domain　coherent　synthesis.　The　results　show　that　the　Q-learning　algorithm　is　easier　to　debug　and　has　better　stability.