Application of Time Serial Model in Water Quality Predicting

Water resources are an indispensable and valuable resource for human survival development. quality predicting plays important role in the protection development of water resources. It is difficult to predict due its random trend changes. Therefore, a method which combines Auto Regressive Integrated Moving Average (ARIMA) clustering model was proposed this paper. By taking monitoring data certain river basin as sample, Total Phosphorus (TP) index selected prediction object. Firstly, sample cleaned, stationary analyzed, white noise analyzed. Secondly, appropriate parameters were according Bayesian Information Criterion (BIC) principle, component characteristics obtained by using ARIMA conduct predicting. Thirdly, relationship between precipitation TP field analyzed K-means method, incremental on changes calculated. Finally, combining with characteristics, results Compared experiments showed that has higher accuracy, Mean Absolute Error (MAE), Square (MSE), Percentage (MAPE) respectively reduced 44.6%, 56.8%, 45.8%.