Modelling Air Pollution Crises Using Multi-agent Simulation

TL;DR

This paper presents a multi-agent simulation combining Gaussian plume and neural network models to evaluate air pollution control strategies during crises, accounting for both controlled and uncontrolled pollution sources.

Contribution

It introduces an integrated multi-agent system that models pollution sources, control agencies, and natural leaks to assess policy effectiveness during air pollution crises.

Findings

Cooperation strategies reduce pollution levels effectively.

Simulation shows the impact of uncontrolled sources on pollution.

Policy assessments help optimize crisis management.

Abstract

This paper describes an agent based approach for simulating the control of an air pollution crisis. A Gaussian Plum air pollution dispersion model (GPD) is combined with an Artificial Neural Network (ANN) to predict the concentration levels of three different air pollutants. The two models (GPM and ANN) are integrated with a MAS (multi-agent system). The MAS models pollutant sources controllers and air pollution monitoring agencies as software agents. The population of agents cooperates with each other in order to reduce their emissions and control the air pollution. Leaks or natural sources of pollution are modelled as uncontrolled sources. A cooperation strategy is simulated and its impact on air pollution evolution is assessed and compared. The simulation scenario is built using data about Annaba (a city in NorthEast Algeria). The simulation helps to compare and assess the efficiency of policies to control air pollution during crises, and takes in to account uncontrolled sources.