Abstract: Lattice-generation is an ideal method for generating lightweight and high-stiffness thin plates, with great potential to significantly reduce material consumption and improve product performance. To lattice thin plates quickly, effectively, and universally into lightweight thin plates that meet stiffness requirements, a smart generation approach of dual-resolution lattice structure is proposed. Firstly, for thin plates with a given standard shape (rectangular cross-section), a series of dual-resolution lattice shaped thin plates are constructed using different ratios of dual-resolution lattice units and a uniform random sampling layout. Secondly, finite element method is used to perform mechanical analysis on them and calculate the relevant stiffness. Thirdly, the above data are collected and construct a stiffness prediction model based on an Artificial Neural Network (ANN), which accurately predict the stiffness of a given thin plates under different dual-resolution lattice structures (with complex nonlinear relationships among them). Finally, a mathematical planning problem is constructed with the goal of generating thin plates with dual-resolution lattices that satisfy the stiffness requirements. The above problem is solved by incorporating the above stiffness prediction model in the particle swarm optimization algorithm to obtain an optimized dual-resolution lattice configuration. The experimental results show that the approach can generate lightweight thin plates with the required stiffness in a fast, efficient, and general way.