In this study, the effects of wavy collecting plate geometry on the performance of a two-stage electrostatic precipitator were numerically investigated. The results showed that variations in plate wavelength primarily affected the local electric field and flow structures, whereas increasing the wave amplitude significantly enhanced particle collection by strengthening the electric field and promoting transverse particle migration. The volumetric flow rate was also identified as a dominant parameter because lower flow rates increased particle residence time and charging, thereby enhancing collection efficiency. Among the representative configurations analyzed, the highest collection efficiency of 73.5% was obtained for TS-Wavy-E at a volumetric flow rate of 0.05 m3/s. An artificial neural network model was further developed to predict collection efficiency based on the geometric and operating parameters. The ANN showed good agreement with the numerical results and outperformed the conventional regression model within the investigated parameter range. These results demonstrate that appropriate modification of the collecting plate geometry can effectively improve particle collection performance and that the ANN provides a useful data-driven representation of the numerical results for two-stage ESP design.