The Regression Model of NOx Emission in a Real Driving Diesel Vehicle

EasyChair Preprint 3419

Abstract

Abstract. The purpose of this study is to use ANN method to train model and predict the NOx emissions of real driving diesel vehicles. In the study, there are two PEMS measurements of driving vehicle’s NOx emission data. One of them was divided into three data types: urban, suburbs and highway according to the driving speed. The data such as vehicle speed, vehicle acceleration and EGR are selected as the features values, and the NOx emissions are the target value. Keras and ANN nonlinear autoregressive exogenous model (NARX) are used to train model and predict NOx emissions values. And then make a comparison with the PEMS measurements. The Keras model performances R^2 of first data are 0.9866, 0.9955 and 0.9962 with respect to urban, suburbs and highway types. And The NARX model performance R^2 of second data is 0.995.

Keyphrases: ANN, EGR, Keras, NARX, NOx, PEMS

@booklet{EasyChair:3419,
  author    = {Hung-Ta Wen and Kuo- Chien Liao and Jau-Huai Lu and Deng-Siang Jhang},
  title     = {The Regression Model of NOx Emission in a Real Driving Diesel Vehicle},
  howpublished = {EasyChair Preprint 3419},
  year      = {EasyChair, 2020}}