Machine learning assisted predictive urban digital twin for intelligent monitoring of air quality index for smart city environment

Abstract

Environmental factors such as urban air pollutants have detrimental effects on human health. In this research a digital twin (DT) based innovative strategy is presented for accurately forecasting Air Quality Index (AQI) in smart city environment. The historic data of Delhi city is collected, and six different deep learning algorithms are implemented to forecast AQI. The 3D model of the smart city is developed in the Blender, and its urban DT is developed in Microsoft Azure. The InfluxDB database is used for storage and retrieval of time-series data. The experimental results show that the CNN-1D-2 layer model outperforms all other algorithms with MAPE of 0.01231, MSLE of 0.00036, R² score reaching 0.99951, and model accuracy of 97.950647. The 3D urban DT model highlights the polluted areas with different colors based on AQI thresholds and DT Grafana dashboard displays the graphical values of AQI and different pollutants along with their trends.