Air pollution modeling based on technical inspection data, traffic congestion, an‎d wind speed using machine learning models

11/4/2026 12:00:00 AM

Air pollution, as one of the serious environmental challenges, has significant negative impacts on public health an‎d quality of life. Accurate prediction of air pollution can contribute to effective policymaking, urban planning, an‎d optimal resource management. This thesis examines an‎d models air pollution using technical inspection data, traffic congestion index, an‎d wind speed. The main objective of this research is to analyze the impact of various factors on air pollution an‎d provide reliable predictions for carbon monoxide (CO), particulate matter (PM10), an‎d sulfur dioxide (SO2). In the first stage, data from the Central Headquarters of Tehran Vehicle Technical Inspection were used as input in long-term modeling. In order to predict pollutants, two models, ARIMA an‎d XGBoost, were used. The results showed that the XGBoost model performed better than the ARIMA model with acceptable values of validation indices; But the significance of CO an‎d HC variables was not confirmed in the technical inspection data. In the second stage, modeling was performed using the traffic congestion index an‎d wind speed. By analyzing the congestion data collected hourly an‎d using the XGBoost, artificial neural networks (ANN) an‎d GRU models, accurate predictions for pollutants were provided. The results showed that air pollution can be effectively modeled using the traffic congestion index an‎d the XGBoost model had the best performance in this field. This research shows the importance of using traffic data an‎d weather changes in air pollution analysis an‎d the high potential of machine learning models in predicting pollutants. Based on the results obtained, the XGBoost model is able to predict air pollution for a pollution measurement station in terms of traffic congestion parameters around the station an‎d the wind conditions in the city in the short term with an R2 index of 0.56, 0.64 an‎d 0.63 for SO2, CO an‎d PM10 pollutants, respectively.

پيش‌بيني آلودگي هوا، شاخص ازدحام ترافيك، معاينه فني خودرو، XGBoost، GRU، NN، ARIMA، داده هواشناسي

Air pollution prediction, traffic congestion index, vehicle technical inspection, XGBoost, GRU, NN, ARIMA, meteorological data

maryam niazi rad

dr reza golshan khavas