Data-driven models for microscopic vehicle emissions. (November 2019)

• Record Type: Journal Article
• Authors: Hajmohammadi, Hajar; Marra, Giampiero; Heydecker, Benjamin
• Is Part Of: Transportation research. Volume 76(2019)
• Page Start: 138
• Page End: 154
• Publication Date: 2019-11
• Subjects: Vehicle emission modelling -- GAMLSS approach -- Air pollution; Transportation -- Research -- Periodicals; Transportation -- Environmental aspects -- Periodicals; 354.76
• DOI: 10.1016/j.trd.2019.09.013
• Languages: English
• ISSNs: 1361-9209

Abstract

Highlights: Emission is described by the spline functions of explanatory variables. Flexible two and three-parameter distributions are used for error distributions. Estimation is improved substantially compared to the GLM and classified regression. Abstract: In this paper, a new approach for describing the relationship between tailpipe emissions and vehicle movement variables is presented, called generalized additive model for location, scale and shape (GAMLSS). The dataset for this model is second-by-second emission laboratory measurements, following a real driving cycle that were recorded in urban, suburban and motorway areas of London. The GAMLSS emission model estimates each of CO2, CO and NO x in each second for two different vehicle types (petrol or diesel) using instantaneous speed and acceleration as the explanatory variables. Comparing the results with current emission models indicates substantial improvement in accuracy and quality of estimation by this approach.