In traditional structural crane design, the fatigue strength of the structure is qualitatively evaluated, only depending on the steel type, working level and joint form, which leads to the excessive service life of some parts of the existing main girder structure, resulting in a waste of industrial resources and manufacturing costs. Based on the concept of equal life design, this paper obtained the service information of the crane in the scheduled inspection period through the rapid prediction method of “field collection+machine learning.” Combined with the bearing capacity evaluation method and the fatigue life assessment method of “S-N curve+fracture mechanics,” a nonlinear mixed integer equal life optimization model of crane girder structure was established, and the multi-specular reflection optimization algorithm was used to optimize the life distribution at different positions and overall weight of the main girder. The structural design parameters of the crane main girder which meet the bearing capacity and have the characteristics of “equal life+lightweight” were obtained. The performance of the optimized main girder structure was verified by finite element simulation. The results show that this method can realize the quantitative control of the life distribution of each position of the crane main girder structure and the lightweight design of the whole structure, and provides a scientific reference for the equal life and lightweight design of large construction machinery products.