Air Pollution Forecasting using Integrated Weather Stations and Convolutional Neural Network (CNN) Algorithm

The main challenge in forecasting pollution levels is the need to consider local weather conditions. Integrating weather data with pollution data is crucial for improving the accuracy of air pollution level forecasts. This research integrates weather data and pollution data to forecast air pollution levels using Convolutional Neural Network (CNN), a deep learning method that can train systems on large datasets by integrating feature extraction and classification processes. CNN is particularly effective in processing data with a grid structure, such as two-dimensional images, and can also handle high-dimensional data, such as video. This study applies an effective CNN method to forecast air pollution levels using data from an 10T-based integrated weather station. Out of 10 experiments conducted using the Convolutional Neural Network (CNN) algorithm, 9 successfully predicted air pollution levels with high accuracy. With RMSE of 1.2639, MAE of 1.2637, and MAPE of 0.63583%, the CNN model performed best in this investigation, exhibiting extremely low prediction errors. Additionally, an R-Square value of 0.98819 indicates that the model can explain almost all the variability in the data.