Clearing the Air: A Multi-Objective Approach to Industrial Air Pollution Prediction with Bi-Directional Stacked LSTM Model

Abstract

Industrial air pollution stands out as a pressing contemporary issue. Prolonged exposure to air pollutants significantly threatens public health, leading to severe respiratory and lung disorders. Currently, the regulations governing the monitoring and control of industrial pollution lack the necessary stringency. Therefore, this study encompasses parameters related to ambient air and stack emissions 1 from industries namely: Clearing the Air: A M ulti-Objective Approach to Industrial Air Pollution Prediction with Bi-Directional Stacked LSTM Model by employing Multi-Objective Bi-directional Stacked LSTM Model (MOBiLe). This study aims to predict emission rates and conducted a comparative analysis among the state-of-art models. The accuracy of the MOBiLe model is assessed by measuring the Mean Square Error and and Validation Mean Absolute Error. Notably, the MOBiLe model exhibited the lowest error among all state-of-art models at 30%, 40%, 45%, 50%, and 60%, respectively. Subsequently, air dispersion model like Gaussian Dispersion Model are utilized on the forecasted emission rates to assess the spreading of air pollutant concentrations originating from the source, specifically at the stack level.