To improve the original design flaws of air duct of air-assisted boom sprayer, a multi-objective optimization approach integrated relevance vector machines (RVM), multi-objective genetic algorithms (MOGA) and fuzzy system were presented. First of all, the multi-objectives of the air duct were constructed to improve the capability of reduced spray drift and increased penetration. The computational fluid dynamics (CFD) analysis of air-flow generating duct were utilized for sampling scheme given by uniform design to collect the train dataset. Sequentially, RVM based on meta-model as fitness function was combined with MOGA to obtain the Pareto optimal set. Finally, a fuzzy inference system was established as decision-making support to obtain the optimum preference solution. Therefore, the optimized air duct structure with the round solution analyzed by CFD showed the promising improvement on flow speed variation. The modified physical prototype proofed feasibility and efficiency of this approach.