Coupled exponential smoothing and gray model for water quality prediction in the Guo River, China.

Abstract

To address the issue of poor prediction accuracy and performance caused by the influence of the original data sequence on the first-order single-variable gray model (GM(1,1)), this study proposes an exponential smoothing gray model (ESGM(1,1)). Taking the Anliu Station situated at the border between Henan and Anhui provinces as an example, ammonia nitrogen and the permanganate index were selected for water quality prediction using the GM(1,1) and ESGM(1,1) models from 2010 to 2021. The fitting accuracy of these models is evaluated by comparing the computed values with the actual monitored water quality index values. The results reveal that the average relative percentage error in the simulation period decreased by 3.01% compared with GM(1,1) and further decreased by 27.41% during the verification period. The mean square error ratio C of GM(1,1) was 0.79, which failed the fitting accuracy test. The C value of ESGM(1,1) was 0.59, which successfully passed the test. The predicted results were consistent with the monitoring data from 2010 to 2021. It is concluded that ESGM(1,1) shows superior accuracy for short-term water quality prediction. This model mitigates the impact of the initial sequence on prediction accuracy and can be utilized for local water pollution control and environmental protection.