Short-Term PM2.5 Prediction Based on Multi-Modal Meteorological Data for Consumer-Grade Meteorological Electronic Systems

Air pollution poses a significant challenge to social sustainability, and accurately predicting PM2.5 concentrations is vital for effective air quality management. In this paper, we introduce a consumer-grade smart PM2.5 prediction system utilizing IoT communication among electronic consumer products. The system incorporates our innovative deep learning model (WVPBL), which integrates wavelet denoising, variational mode decomposition, and principal component analysis to extract features from multi-modal air quality data for short-term PM2.5 concentration prediction. The bidirectional long-short memory network (BiLSTM) model is employed for accurate prediction, considering the nonlinear and dynamic characteristics of PM2.5. The performance of our WVPBL fusion model is assessed using ten sets of air quality data, evaluating mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R^{2}) . Experimental results demonstrate superior prediction performance compared to other models, with the highest RMSE at 1.8247.~R^{2} remains consistently high. The findings emphasize the importance of integrating deep learning and social computing in PM2.5 prediction, enhancing the functionality of electronic consumer products, and fostering sustainable social development.