Roadside vehicle particulate matter concentration estimation using artificial neural network model in Addis Ababa, Ethiopia

Currently, vehicle-related particulate matter is the main determinant air pollution in urban environment. This study was designed to investigate level of fine (PM2.5) and coarse particle (PM10) concentration roadside vehicles Addis Ababa, capital city Ethiopia using artificial neural network model. To train, test validate model, traffic volume, weather data concentrations were collected from 15 different sites city. The experimental results showed that average 24-hr PM2.5 13%–144% 58%–241% higher than quality index (AQI) world health organization (WHO) standards, respectively. PM10 also exceeded AQI (54%–65%) WHO (8%–395%) standards. model runs Levenberg-Marquardt (Trainlm) Scaled Conjugate Gradient (Trainscg) comparison performed, identify minimum fractional error between observed predicted value. two models determined correlation coefficient other statistical parameters. Trainscg exhaust emission be (R2 = 0.775) 0.92), Trainlm has well 0.943) 0.959). overall a better obtained could considered as optional methods predict transport-related volume for cities countries have similar geographical development settings.