6061 Aluminum alloy has low density, good corrosion resistance, oxidation resistance and weldability, therefore, is widely used in aerospace field. However, 6061 aluminum alloy has poor formability at room temperature, so hot forming technology is normally used to improve the quality of its thin-walled parts. The hot forming process of a double-C thin-walled part of 6061 aluminum alloy was studied in this paper; uniaxial hot tensile tests were conducted to investigate the deformation behavior of 6061 aluminum alloy under different temperatures and strain rates. By comprehensively considering the coupled effects of temperature and strain rate on forming quality, an improved Johnson–Cook (J–C) constitutive model was proposed to describe flow stress of materials, then parameters of the improved constitutive model were characterized using genetic algorithm. A finite element model of nonisothermal forming for the double-C thin-walled part was established, and orthogonal experiments and range analysis were conducted to rank the influence of various process parameters on stamping quality of the double-C thin-walled part. Latin hypercube sampling was employed to obtain training samples, and test samples were randomly generated. The maximum thinning rate of the double-C thin-walled part was taken as the optimization objective and simulation by ABAQUS was used to obtain response values for different samples, then an improved BP neural network was utilized to establish a mapping relationship between process parameters and the maximum thinning rate. The optimal combination of process parameters was obtained through an improved genetic algorithm, and the effectiveness of this method was verified through experiments.