Air-quality prediction based on the ARIMA-CNN-LSTM combination model optimized by dung beetle optimizer

Area of Science:

• Environmental Science
• Data Science
• Artificial Intelligence

Background:

• Air pollution poses significant threats to economic development and public health.
• Efficient and accurate air quality prediction models are crucial for effective pollution management.

Purpose of the Study:

• To develop a superior combined model for predicting Air Quality Index (AQI).
• To enhance the accuracy of air quality forecasting for better pollution management.

Main Methods:

• A hybrid model combining ARIMA for linear data patterns and CNN-LSTM for non-linear patterns was developed.
• The Dung Beetle Optimizer algorithm was employed to optimize CNN-LSTM hyperparameters, addressing hyperparameter setting challenges.
• The proposed model was validated against real AQI data from four cities and compared with nine other models.

Main Results:

• The proposed combined model demonstrated superior performance across all metrics: root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R²).
• Specific RMSE values achieved were 7.594, 14.94, 7.841, and 5.496 for the four cities.
• MAE values were 5.285, 10.839, 5.12, and 3.77, with R² values of 0.989, 0.962, 0.953, and 0.953, respectively.

Conclusions:

• The novel combined ARIMA-CNN-LSTM model with Dung Beetle Optimizer significantly outperforms existing models in AQI prediction.
• This advanced model offers a more accurate and reliable tool for air pollution management and public health protection.