Forecasting on PM10 in South Korea Using the CNN‐BiGRU‐LSTM Model

After the COVID‐19 pandemic in Korea and worldwide, factory operations have resumed. As environmental pollution accelerates, the number of days with a severe fine‐dust concentration has been rapidly increasing. The higher fine‐dust concentration results in more dust entering the lungs or brain, adversely affecting the body. This study was conducted to guide national or personal alternative methods of prevention when the fine‐dust concentration is severe. If the fine‐dust concentration forecasting system is accurate, high concentrations could be more reliably prevented. Based on the weather and air pollution variables, this study focused on improving predictive performance by comparing the MLP, basic ANN models, RNNs (RNN, LSTM, and GRU), and CNN, which are time‐series DL models, and implementing a new hybrid combination model. As a result, the hybrid models, CNN‐BiGRU and CNN‐BiGRU‐LSTM, improve prediction performance better than the single DL models.