Breathable Cities: Dynamic Machine Learning Modelling Approaches for Advanced Air Pollution Control

Aug 5, 2026

Upload by Hazel

This paper addresses air quality index (AQI) representation in ambiguous boundary regions using a fuzzy logic framework to extend air quality prediction practices. It reviews existing air quality prediction (AQP) standards, then proposes a fuzzy air quality levels prediction (FAQLP) model that maps actual AQI ranges and classifies predicted fuzzy levels, aiming for more dynamic cross-country comparability. The study targets uncertainty near AQI boundaries and introduces a hybrid DNN–Markov model for accurate hourly AQI predictions with ANFIS-based air quality level representation.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages20

Reading Time50 min

AccessFree to download

Introduction

Air quality monitoring and public policy relevance

AQI definition and limitations

Motivation for hybrid modeling

Methodology and modeling framework

Fuzzy logic for AQI boundary uncertainty

FAQLP model for air quality level prediction

Hybrid DNN–Markov and ANFIS approach

Comparative urban settings

Dynamic machine learning for two cities

Conclusion

applied sciences Article Breathable Cities: Dynamic Machine Learning Modelling Approaches for Advanced Air Pollution Control Roba Zayed and Maysam Abbod * Citation: Zayed, R.; Abbod, M. Breathable Cities: Dynamic Machine Learning Modelling Approaches for Advanced Air Pollution Control. Appl. Sci. 2024, 14, 5581 . [https://doi.org/10.3390/app14135581](https://doi.org/10.3390/app14135581) Academic Editor: Yves Rybarczyk Received: 20 May 2024 Revised: 19 June 2024 Accepted: 24 June 2024 Published: 27 June 2024 Copyright: © 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ([https://](https://)[ ](https://)[creativecommons.org/licenses/by/](creativecommons.org/licenses/by/)[ ](creativecommons.org/licenses/by/)[4.0/](4.0/)) . Department of Electronic and Electrical Engineering, Brunel University London, London UB8 3PH, UK; [roba.zayed@brunel.ac.uk](roba.zayed@brunel.ac.uk) * [Correspondence: maysam.abbod@brunel.ac.uk](Correspondence: maysam.abbod@brunel.ac.uk) Abstract: This paper discusses air quality index (AQI) representation using a fuzzy logic framework to cover the blurry areas of AQI where indices are in between ranges of values. After studying several standards for air quality prediction (AQP), this research suggested the use of fuzzy logic as an extended method to cover some limitations found in several standards, in which the fuzzy logic represents a more dynamic way to support cross-country comparisons as well. This research expanded upon the United States Environmental Protection Agency (USEPA) standards to address their acknowledged limitations by constructing a fuzzy air quality levels prediction (FAQLP) model, which categorizes air quality into corresponding ranges (actual levels) and classifies new fuzzy levels (predicted levels), using a fuzzy logic model (to enforce more realistic predictions) . This model can solve the issue of values at or near boundaries when there is uncertainty about air quality levels. The study aims to incorporate a comparative study of two urban settings providing dynamic machinelearning modeling approaches for advanced air pollution control. The DNN–Markov model is presented in this paper as the selected hybrid model for AQI prediction, and the adaptive neuro-fuzzy inference system (ANFIS) was used to represent AQI. This work presents a novel air quality index framework that consists of a DNN–Markov model for accurate hourly predictions and air quality level representations using ANFIS. Keywords: air quality index; fuzzy logic; machine learning modeling; prediction; adaptive algorithms; air pollution forecasting; air quality monitoring; artificial intelligence; dynamic modeling; hybrid models 1. Introduction Monitoring air quality is increasingly a necessity given continuously increasing pollution levels, with diverse and serious consequences on human health, social interactions, atmospheric impacts, and ultimately socio-economic development. Artificial intelligence (AI), machine learning (ML), and deep learning uses are advancing over time, which has been leveraged by this study aiming to offer enhanced predictive accuracy and a reliable tool for policymakers. This paper presents advanced ML approaches for air quality monitoring. This research is relevant from different perspectives, given the increasing impact of air quality on public health and policymaking. The methods used have demonstrated approaches for reliable accuracy and operational feasibility in urban settings, leveraging multivariate data to provide hourly forecasts for air quality in two diverse urban environments. While the term ‘air pollution’ reflects the presence of pollutants in the air, the broader concept of ‘air quality’ alludes to the general quality of the air we breathe. Clean air is a very basic need for humanity, for health and other life a

FAQs

How does the paper handle uncertainty in AQI boundary regions?

It uses a fuzzy logic framework to represent “blurry areas” where AQI values fall between standard ranges, improving robustness when uncertainty exists near boundaries.

What models are proposed for AQI prediction and air quality level representation?

The paper presents a DNN–Markov hybrid model for accurate hourly AQI predictions, and uses ANFIS to represent and classify air quality levels.

Why is fuzzy logic used alongside existing air quality prediction standards?

The research extends and addresses acknowledged limitations in existing standards by creating FAQLP, enabling more dynamic support for cross-country comparisons and more realistic predictions.

Aug 5, 2026

Upload by Hazel