Abstract: A286 superalloy sleeves are widely used in aerospace connections components. Two kinds of cold heading forming schemes for sleeves were developed, and the finite element simulation software DEFORM-3D was used to simulate and analyze the cold heading process of the sleeves. The forming effect of the two schemes was compared, and the metal flow law, forming load variation, metal flow line distribution and effective strain characteristics during cold heading were analyzed thoroughly. Aiming at the forming load of the head of the sleeve, the optimization research was carried out. A neural network model was developed to predict and optimize the maximum forming load, and the optimal combination of process parameters was determined. The results show that the optimal process parameters are as follows: forming speed of 30.45 mm/s, friction factor of 0.081, pin angle of 80.47°, and pin rounding angle of 0.57 mm. The error between the predicted data of the neural network and the simulation results is less than 3.6%. The research results can provide a reference for the cold heading process design of thin-walled sleeve components.