Integrated air quality prediction with seasonal index and grey model

Abstract

As the capital of China, Beijing's air quality monitoring and prediction are the focus of social attention. Through the analysis of the air quality index (AQI) in Beijing over the past decade, the changes in air quality are influenced by various uncertain factors, presenting complex and diverse characteristics, with obvious nonlinear and seasonal features, and a long-term downward trend. To this end, an improved second-order adaptive grey prediction model was proposed to predict annual AQI, and combined with seasonal index (SI) to predict monthly AQI, forming an Air Quality Index Prediction System (AQIFS). Compared to traditional grey models, the new model introduces dynamic power functions to optimize background values, thereby better fitting the nonlinear trends in the sequence. And it combines SI decomposition with monthly data to accurately capture seasonal fluctuations. Experimental results show that the proposed model performs the best in terms of annual average relative percentage error (MRPE) are 2.98 % and 3.82 %, root mean square error (RMSE) are 3.3593 and 3.4244, respectively. Its comprehensive performance is significantly better than other models. This study has made a positive contribution to enriching and improving the methodological system of grey prediction models, providing an effective modeling method for predicting seasonal fluctuations in air quality.