Modeling the relationship between air quality and intelligent transportation system (ITS) with artificial neural networks.

Air quality;Intelligent transportation system;Artificial neural networks

Dinesh Kumar Gupta, 1965-

University:University of Louisville

Department:Civil and Environmental Engineering

关键词: Air quality;    Intelligent transportation system;    Artificial neural networks;   

Others  :  https://ir.library.louisville.edu/cgi/viewcontent.cgi?article=1548&context=etd

美国|英语

来源: The Universite of Louisville's Institutional Repository

【 摘 要 】

Environmental or air quality impacts of Intelligent Transportation Systems (ITS)are very difficult to measure. Some researchers have attempted to quantify the effects ofindividual ITS application on emissions; yet, the effects of ITS as a whole on ambient airquality have not been investigated.The objective of this research was to model the relationship between ITS andambient air quality. The multiple Artificial Neural Networks (ANN) training with thedata yielded a model for predicting the air quality. In addition, the ANN made themeasurement of the effect of ITS on air quality possible.Data pertaining to sixty US cities (urbanized area) were used for this research.Input variables used were related to transportation and local characteristics, and ITSapplications. Output variables were the annual average concentrations of CO, Ozone, andN02 in ambient air. The K-fold cross validation technique was used to train the ANN.The results of ANN model were compared with that of a Multiple Regression (MR)model showing the supremacy of ANN over MR. The ANN model results show that theMean Absolute Errors (MAEs) in prediction vary from 5 to 20 %. This variance isjustified since the factors related with industries, which contribute significantly to air pollution, have not been taken into consideration in this study. There were some unusualfindings: in contrast to the common assumptions, N02 concentration increases with ITSintensity, and Ground Level Ozone concentration, in ambient air, seemed to be moretransportation-dependent as compared with that of CO and N02•A recommendation for further research on this topic is to include more inputvariables, especially those which are relatcd with industries, to improve the accuracy ofprediction. Scientific experimentations have also been recommended to corroborate theunusual findings.