The miniaturization of the scrubber used in vessel flue gas desulphurization is a critical issue. In this study, the structure optimization of the vessel desulphurization scrubber based on CFD (computational fluid dynamics) and the SVM‐GA (support vector machine and genetic algorithm) is performed and a miniaturized scrubber design with the same desulphurization efficiency as large scrubbers is obtained. First, the seawater SO2 absorption process in a vessel desulphurization scrubber is investigated using the CFD‐DPM method, and the effects of the operating and structural parameters on the desulphurization efficiency η are discussed in detail. Results show that η is positively correlated to the absorption area height, H2, and the spray level number, N, and η first increases and then decreases with an increase in the scrubber diameter, D1, as well as the inlet flue angle, θ. Then, a prediction model of η considering D1, H2, θ, and N is established and validated using the SVM with the simulation data. Finally, D1, H2, θ, and N are optimized using the GA method based on the SVM model. The results show that the volume of the absorption area, Vab, can be reduced by 30 % while maintaining the same η through the use of this optimization method.