The work aims to obtain the optimal process control variables for the pulp refining process to ensure the economical production with superior beating quality, high yield and low cost. With flow, concentration and disc refiner's current as decision variables, the multi-objective optimization model was established by regarding the stability of beating degree and wet weight, the amount of dry weight of beaten pulp in unit time and the refining cost per unit as the modeling indexes, an improved dynamic adaptive simulated annealing algorithm (DASAA) was proposed and the simulation solution was optimized based on Matlab. The case simulation results showed that the actual effects before and after the optimization were significant; in theory, not only the refining quality was better, but the yield was increased by 0.73 t/h and the cost was reduced by about 4.25 yuan/t. Compared with the empirical setting control of traditional decision variable, the optimization control strategy of serial refining process based on the improved optimization algorithm can better meet the production requirements, and to some extent achieve the trade-off optimization between high quality, high yield and low cost.