Air Pollution Forecasting Using LSTM-Multivariate Regression Model

Abstract

There are two kinds of air pollutants, such as primary and secondary. Primary pollutants are emitted straight by vehicles such as CO, CO₂, SO₂, NO, NH₃, NO₂, PM10 and PM2.5. Secondary pollutants happen when communicating with each other in the environment. Atmospheric particles or particles, including carbon, sulfur, nitrogen and metal compounds, may be small components or liquid in the environment and consist of hundreds of separate chemicals. Researchers use different machine learning algorithms and struggle to get PM10 and PM2.5 more precise. In this paper, we suggest a regression model for LSTM/Multivariate Variate to predict the more precise PM2.5 value during summer and cold sessions. Finally, the LSTM/MVR model is compared to the LSTM and the outcome demonstrates that the suggested technique efficiently predicts a next one-hour PM2.5 mistake relative to the LSTM error.