Multi-model Ensemble Air Quality Prediction with Dung Beetle Optimization

Air quality prediction is challenging because it is affected by a multitude of complex factors such as meteorological conditions, anthropogenic emissions, and geographical location, which traditional single models often struggle to fully capture in their dynamic interactions, limiting the accuracy and reliability of predictions. To address this issue, a Dung Beetle Optimization based multi-model ensemble air quality prediction method (MMEDBO) is proposed in this paper. This method adopts three advanced deep learning models: Convolutional Long Short-Term Memory network, Convolutional -Gated Recurrent Unit network, and Temporal Convolutional Network. Utilizing the DBO algorithm, the weights of each model are automatically determined to optimize prediction performance. Compared to single model prediction methods, MAE and MSE have decreased by more than 0.54 and 0.79, respectively. Experimental results demonstrate that, compared to single models, MME-DBO exhibits higher accuracy and robustness, proving its efficacy in complex environmental data analysis.