Abstract: The leaf spring is an important part of a nuclear fuel assembly. Its performance is directly related to the service safety of fuel assemblies. Considering the complex non-linear factors of leaf spring materials, such as the elastic-plastic constitutive relation and large deformation, the authors, using the ANSYS code, performs the parametric modeling of leaf spring under the constraint of multi-scale coupling. The model geometry, hexahedral mesh and contact pairs of the leaf spring are established automatically. On this basis, the authors study and analyze the influence of key parameters on the characteristics of leaf springs. Depending on the MATLAB parallel computing library, the authors build a multi-parameter optimization platform for leaf spring characteristics based on intelligent particle swarm optimization (IPSO) algorithm. Then, the authors use this platform to optimize the leaf thickness, variable cross-section position and arc-shaped transition zone form over the design stiffness curve and minimum plastic deformation of the leaf spring. As demonstrated by the results, the intelligent multi-parameter optimization algorithm based on particle swarm can help improve the leaf spring design efficiency greatly; and within the given leaf spring design target curve and the leaf spring parameters, this algorithm can provide structural parameters satisfying the design target via a less times of iteration, which provide a good guidance on the nuclear reactor leaf spring engineering.