Air quality analysis and prediction are very important in environmental research as airborne pollution has become a significant health threat, especially in Chinese urban agglomerations. Most previous analysis systems have been based on direct factors, such as pollutant concentrations, wind speeds and direction, relative humidity, and temperature; however, the air quality in a city is also affected by the air quality conditions in surrounding areas. This paper proposes a novel strategy for the analysis and forecast of air quality levels, for which Artificial Neural Networks (ANNs) are employed to elucidate the complex relationships between air quality and meteorological predictor variables. The experimental results in the study demonstrated that the normalized EEMD-ANN model outperformed other models in terms of the Precise, MAE and MAPE. The proposed model, therefore, demonstrated its potential as an administrative tool for issuing air pollution forecasts and for designing suitable abatement strategies.