Analysis of NOx, HC, and CO Emission Prediction in Internal Combustion Engines by Statistical Regression and ANOVA Methods

Abstract

This study investigates the impact of engine operating parameters on exhaust emissions, mainly nitrogen oxides (NOx), hydrocarbons (HC), and carbon monoxide (CO), using a statistical regression modelling approach. The analysis employs the Analysis of Variance (ANOVA) method to assess model accuracy and sensitivity. The results indicate that the developed regression models exhibit high predictive accuracy, with the CO model demonstrating the best performance. The NOx model achieved an R-squared value of 0.9827, explaining 98.27% of data variation, but showed the highest standard deviation (35.14) and PRESS value (1.549E+005), indicating more significant data variability. The HC model performed slightly better with an R-squared value of 0.9865, a standard deviation of 3.68, and the lowest coefficient of variation (C.V.%) at 1.68%, ensuring high stability. The CO model outperformed both, with the highest R-squared value (0.9910), the lowest standard deviation (0.11), and a moderate C.V.% of 3.70%, making it the most reliable. Additionally, the Adeq Precision values for NOx, HC, and CO models were 45.791, 51.764, and 66.569, respectively, confirming strong signal-to-noise ratios. The findings indicate that increasing engine speed and throttle opening significantly influences emissions. Higher speeds increase NOx emissions but reduce HC and CO emissions, while larger throttle openings increase CO levels. These results provide valuable insights into optimizing engine parameters for reducing emissions. The developed models can be practical tools for emission predictions and mitigation strategies, contributing to environmentally friendly combustion technologies.