The sugar factory's sugar crystallization is the last and the most crucial step. It incorporates both the heat and mass transfer. Controlling the supersaturation stability is the key factor that impacts the boiling sugar in the crystallization process. As a result, we must keep supersaturation in metastable zone (keeping the supersaturation in 1.20-1.25) in order to speed up the crystallization rate and ensure the quality of crystal sugar. So how to ensure the stability of the supersaturation is the key to automation in sugar boiling process. Adaptive dynamic programming can solve nonlinear and long-horizon optimization problem not have a precise mathematical model of the plant. In this paper, based on a large number of data about brix coming from the actual boiling sugar process in the factory, action dependent heuristic dynamic programming is used to optimize the sugar crystallization process by controlling the brix of the sugar solution. The result of simulation shows this controller which is designed by using the action dependent heuristic dynamic programming method can optimize the sugar crystallization process.