Air Quality Index Forecasting in the HRcity Smart City System Based on an LSTM Prediction Model

Abstract

The HRcity project aims to implement various digital technologies for the needs of cities in the Republic of Croatia. The system itself consists of several different components specialized for the needs of citizens and various institutions operating in the city areas. One of these components monitors air quality and measures the Air Quality Index (AQI). This paper presents the structure of the AQI component and the principles for measuring and monitoring pollution values and determining the AQI. Air quality prediction models, based on LSTM, were implemented based on pollutant measurements collected during three years of system use. RMSE, MAE, and MAPE metrics were used to evaluate the performance of the LSTM models. The experimental evaluation shows that the LSTM models implemented with other elements of the AQI achieved very good accuracy.