Air Quality Index Prediction Model Based on Multiple Attention Mechanisms and Hyperparameter Optimization

Air pollution poses extensive threats to human health and the environment. Air quality forecasting helps people take preventive measures in advance, reducing unnecessary losses. This paper introduces a new Air Quality Index(AQI) prediction model named SSA-CCBA, which is based on a Convolutional Neural Network Bidirectional Long Short-Term Memory network (CNN-BiLSTM) foundational model, combined with multiple attention mechanisms and hyperparameter optimization. Firstly, on a conventional CNN network, the Convolutional Block Attention Module (CBAM) is incorporated to enhance the network's feature extraction capability. Subsequently, an attention mechanism is introduced at the output of the BiLSTM network to enhance the temporal features extracted by BiLSTM, reinforcing the influence of crucial historical information. Lastly, the Sparrow Search Algorithm (SSA) is employed for hyperparameter optimization, selecting the optimal hyperparameters to unleash the model's best performance. Simultaneously, experimental validation was conducted on real-world datasets, demonstrating that the model exhibits higher prediction accuracy.