Mihaela OPREA, Elia Georgiana DRAGOMIR, Marian POPESCU, Sanda Florentina MIHALACHE<br>Particulate Matter Air Pollutants Forecasting using Inductive Learning Approach

Mihaela OPREA, Elia Georgiana DRAGOMIR, Marian POPESCU, Sanda Florentina MIHALACHE
Particulate Matter Air Pollutants Forecasting using Inductive Learning Approach

Abstract:

Particulate matter (PM) represents an important air pollutant with potential significant negative effects on human health when the level of concentration exceeds certain limits, imposed by national and international air quality standards. More accurate forecasting of such levels of PM concentrations became an important challenge nowadays, for the environmental protection specialists. Depending on the pollution sources, PM air pollutants have a variety of chemical and physical compositions. Moreover, the meteorological and geographical conditions as well as the existence of other air pollutants in the same region (to which PM can interact or which generate PM via chemical reactions), will influence the real levels of PM concentrations. Due to the high complexity of physical or chemical models that provide a more complete characterization of the PM related air pollution trends, other approximate models could be adopted. An example is an artificial intelligence based forecasting model that incorporates knowledge from the environmental expertise domain, which will guide the forecasting process. The research study presented in this paper proposes such a model, based on a machine learning approach, inductive learning, that extracts rules for guiding forecasting of the PM air pollutants concentrations levels, with better accuracy. A comparative analysis between two forecasting models based on the inductive learning algorithms, REPTree and M5P, was carried out for the forecasting of the next day PM10 (i.e. PM with the diameter less than 10mm) concentration level by using the last 8 days measured values. The experiments that were performed revealed that the M5P inductive learning algorithm improved the accuracy of the short-term PM10 concentrations levels forecasting. Keywords: environmental protection, particulate matter air pollutant forecasting, artificial intelligence, inductive learning

Issue: 2016, Volume 67, Issue 10

Pages:

Download Pdf Article

This article is published under the Creative Commons Attribution 4.0 International License

<<<< back

PUBLISHING SECTION

Author Guidelines

* Paper Drafting

* Paper Template

* Ethical Statement

* Paper Submission

Ghidul Autorului (ro)

Open Access & Archiving Policy

Search Authors

LATEST NEWS
	NUMAR OMAGIAL PETRU PONI.. detalii
	Revista de Chimie - 70 de ani de aparitie neintrerupta.. detalii
	English-Romanian Dictionary for Mechanical Engineering (download pdf)