Previous studies have demonstrated that that facade ribs serve as an effective aerodynamic optimization measure, significantly reducing wind loads on high-rise buildings. However, the optimal arrangement of facade ribs under minimal wind loads in practical engineering remains unclear. Conventional numerical simulations and wind tunnel test methods can only obtain limited optimization schemes for rib layouts, and the optimization process is time-consuming and labor-intensive. Therefore, this paper utilizes a multi-objective optimization procedure combining Large Eddy Simulation (LES), Back Propagation Neural Network (BPNN), and Non-Dominated Sorting Genetic Algorithm (NSGA-II) to assess the optimal layout parameters for facade ribs corresponding to minimal wind loads. The results demonstrate that the BPNN surrogate model can rapidly capture the complex nonlinear relationship between rib layout parameters and target wind loads, revealing notable differences in the load variation trends on the overall force on the model and on ribs positioned at different locations. The mean drag and fluctuating lift of models of windward ribs and upstream sidewall ribs have opposite trend.  The loads on the windward ribs and upstream side ribs exhibit opposite trends compared to the total lift and drag on the model. Due to the influence of vortex shedding, the loads on the downstream side ribs and leeward ribs are more complex. The genetic algorithm NSGA-II effectively evaluates the optimal trade-off solutions among multiple objective wind loads. The optimal layout parameters, specifically the position ratio b/D and extension length d/D, fall within the ranges of 0.14-0.17 and 0.073-0.077, respectively, with a b/d ratio of approximately 2. These optimal arrangement parameters provide balance for wind loads on both the model and the ribs, offering references for engineers and designers in wind-resistant design using facade ribs.