SO2 in Atmosphere Predicted with Improved Error GM (1,1) Model-Based on Optimization of Initial Condition in Chongqing, China

摘要

To increase the prediction precision of GM (1,1) model, optimization of the initial condition and error GM (1,1) model was integrated for the improvement of original GM (1,1) model. The results of numerical example indicated that the original GM (1,1) model and the improved GM (1,1) model could mostly indicate the average change tendency of reported value and the error GM (1,1) model and the improved error GM (1,1) model both tended to the actual numerical fluctuation. There were significant correlations among predicted value from four GM (1,1) models and reported value and the correlation was 0.953, 0.959, 0.980 and 0.992, respectively. Taking into account the results of correlation analyses, 0.040 to 0.041 mg/L was considered to be the most reasonable predicted concentration range to SO2 in atmosphere environment of Chongqing, China in 2011. Although the new modified model could improve the prediction accuracy of GM (1,1) model, which was recommended for a small amount of information modeling and prediction, only limited numerical example could be predicted, mainly due to residual error increased in the model.