At present, the flue aimed temperature in coke oven production is mainly determined by artificial experience, which leads to the lack of theoretical guidance in the control process, and the parameters are difficult to adjust in real time. In this paper, the optimization model of the flue aimed temperature of coke oven is established. The correlation model between coke quality, coke yield, coking energy consumption and the flue aimed temperature is analyzed. Firstly, the factors affecting coke quality, coke yield and coking energy consumption are determined by mechanism analysis. Secondly, an improved grey relational analysis method is proposed to quantitatively analyze the correlation between coke quality, coke yield, coking energy consumption and their influencing factors, so as to determine the input of multi-objective optimization model. Because the neural network has a strong nonlinear approximation effect, the RBF neural network is introduced to establish a prediction model of coke quality, coke yield and coking energy consumption. Then, maximizing coke yield and minimizing coking energy consumption are chosen as the optimization object, the coke quality and production boundary are chosen as constraints, and the local optimization targets as decision variables, whereby the multi-objective optimization model is designed. The multi-objective genetic algorithm is applied to solve this multi-objective optimization problem. The optimal value of the flue aimed temperature of coke oven is obtained. Finally, the practical running results show that the proposed multi-objective optimization model has good applicability.