This paper presents an integrated simulation and machine learning (ML) approach for predicting the efficiency of diesel engine aftertreatment systems. A physics-based MATLAB GUI was developed to simulate electrostatic particulate removal under varying operating conditions. The simulation incorporates corona discharge physics, exhaust gas dynamics, and electrostatic field interactions to estimate particle collection efficiency and power consumption. To explore the design space systematically, a synthetic dataset was generated by varying engine and electrostatic precipitator (ESP) parameters. In total, 72,360 simulation runs were conducted, and zero-efficiency cases were removed to ensure data quality. A supervised regression model using the bagging ensemble method implemented in MATLAB’s fitrensemble was trained. The model achieved excellent performance with an R2 of 0.9937 and an RMSE of 1.72. The methodology is scalable and suitable for integration into design tools, optimisation workflows, or real-time estimation systems. Insights gained from the trained model also revealed the most influential parameters, with applied voltage, medium factor, and roughness factor ranking highest. This work lays a foundation for augmenting engineering simulations with ML to accelerate system development and support intelligent control strategies.