Predicting Air Pollution Levels Using LSTM Networks And Particle Swarm Optimization

Abstract

Air pollution is a major contributor to both public health issues and climate change, representing one of the most pressing challenges faced by humanity. Consequently, accurate forecasting of air pollution has become increasingly important. In this study, we propose a predictive model that integrates the Particle Swarm Optimization (PSO) algorithm with a Long Short-Term Memory (LSTM) deep learning framework. The model is designed to optimize the hyperparameters of the LSTM network and forecast the Particulate matter 2.5-micron concentration for the following day using historical Particulate matter 2.5-micron data. The optimization-enhanced model demonstrates superior performance, yielding a lower Root Mean Square Error (RMSE) compared to traditional machine learning approaches. Notably, the proposed model achieves an RMSE of 2.42 in Particulate matter 2.5-micron prediction.