Analyzing vehicle emissions using a hybrid machine learning approach using weighted average based k-means clustering for sustainable transportation decision-making

Abstract

In today’s world, the heavy reliance on crude oil for transportation has significant environmental implications, contributing to air pollution and greenhouse gas emissions, exacerbating global climate change and posing health risks. As environmental sustainability concerns grow, there is a need to explore alternative fuel options and vehicle technologies with reduced emissions. In this study, a comparative analysis is conducted on three distinct vehicle types — hybrid, diesel, and biodiesel — assessing their carbon footprint based on emissions of CO₂, CO, NOx, SO₂, PM and UBHC across various operating conditions such as load, efficiency losses, and torque. The Analytic hierarchy process (AHP) method was used to determine the weights for various output parameters, including, the weights assigned to these parameters are as follows: CO₂ emissions-13.76%, CO emissions-18.29%, UBHC emissions-25.58%, NOx emissions-13.54%, PM emissions-7.01%, and SO₂ emissions-21.83%. Various vehicle types were ranked using the Evaluation Based on Distance from Average Solution (EDAS) approach. The experimental findings show that, out of the three vehicle types, hybrid vehicles had the best emissions profile, with lower levels of all assessed pollutants. Consequently, hybrid vehicles are identified as having the lowest carbon footprint, followed by diesel vehicles, with biodiesel vehicles exhibiting the highest emissions. K-means clustering is used to determine which type of vehicle is most effective at reducing emissions. With emissions of 95 g/km, 0.2 g/km, 0.015 g/km, 0.02 g/km, 0.001 g/km, and 0.005 g/km for CO₂, CO, UBHC, PM, and SO₂, the hybrid car in cluster 1 yields the most promising results. This study underscores the importance of considering environmental impacts in vehicle selection and highlights the potential of hybrid technology in mitigating carbon emissions, highlighted by an insightful K-means clustering study.