Aiming at the problem that the pH value of enzyme-producing strain YS—80 can not be controlled stably in the optimal range during the fermentation of marine protease（ MP）

a pH value control method based on improved iterative learning control（ ILC） is proposed. Firstly

the dynamic model for MP fermentation process is analyzed and the control variables are determined. Then

the BP neural network is trained and tested to obtain the mapping relationship between the gain parameters and control errors of the iterative learning controller. The optimized artificial fish swarm algorithm（ AFSA） is used to realize the dynamic tracking of the gain parameters.Finally

pH iterative learning control model for MP fermentation process is established to solve the problem of YS—80 acid producing curve and unknown growth curve. The simulation results show that the tracking error of this method on matrix concentration and bacterial concentration is less than 0. 669 g/L

and the tracking error of relative enzyme activity is less than 0. 98 %. This control method has good robustness and response speed.