The Rectisol process is widely used for removing acidic gases and purifying hydrogen. However, this process has problems such as high energy consumption and operating costs, which is particularly prominent in the deacidification gas absorbent recovery system of Rectisol devices. Therefore, taking the absorbent recovery system of deacidification gas as the research object, Aspen Plus was used to model the process. Under the feed condition of 80% coke+20% coal, the relative errors between the simulation results and the factory’s actual data for key parameters were less than 5%, which verified the reliability of the model. The mass reflux ratio of the thermal regeneration tower, the top output of the thermal regeneration tower, and the top output of the methanol-water separation tower were selected as the optimization variables, and the annual total utility cost and the H₂S content in the circulating methanol were selected as the optimization objectives. A multi-objective optimization framework based on a combination of Python and Aspen Plus was established, and the non-dominated sorting genetic algorithm Ⅱ(NSGA-Ⅱ) was used to optimize the deacidification gas absorbent recovery system. The method was used to optimize the working conditions for three feed ratios (100% coke, 80% coke+20% coal, 100% coal). The results show that as the proportion of coke in the feed decreases, the H₂S content in the raw material decreases, the mass reflux ratio of the thermal regeneration tower increases, the top output of the thermal regeneration tower decreases, and the top output of the methanol-water separation tower initially increases and then decreases. The total annual utility cost and the H₂S content in the circulating methanol are reduced. The results can provide a basis for changing the operating parameters of the deacidification gas absorbent recovery system when changing the feed ratio.