Prediction of pollutant emission using partial least squares method

In the general context of a Decision support system (DSS), we applied partial least squares (PLS) as multivariate statistical technique for modeling and evaluation of automotive pollutant emissions. During the development of this activity different problems such as the integration of the micro simulated approach of driving cycles in the structure of the DSS have been addressed. This paper presents some results of PLS, related to a case study of emissive data coming from a data set of EURO 4 cars. Also the emission model predicted values on driving cycles, generated by car-following micro simulating models, are evaluated and compared with COPERT IV curves.