Abstract: Investigating fluid flow characteristics during vacuum screening is crucial for revealing its underlying mechanisms. However, the filter bed formed by solid particles retained by the screen during vacuum screening is difficult to describe using microstructures and is often assumed to be a porous media bed. To accurately describe the fluid flow characteristics of the porous media bed and further investigate the performance during vacuum screening, this study predicted the flow resistance coefficient of the porous media bed based on experimental and machine learning methods. The study analyzed the feature importance ranking results of the BP neural network, random forest, and XGBoost models, identifying the sensitivity order of factors affecting the flow resistance coefficient of the porous medium bed as air flow rate > particle bed > particle ratio > particle layer thickness > screen mesh. Ultimately, utilizing the genetic algorithm (GA) with the XGBoost prediction model as the fitness function for multi-objective optimization, a multi-objective optimization model was established, yielding the optimal vacuum screening process parameters and flow resistance coefficient. This study provides a new method for the study of the vacuum screening resistance coefficient, and the results are of great significance for analyzing the vacuum screening mechanism.