Using Clustering Analysis to Predict Outflowing Water Quality of Selectively Drawn Water from Water Sources and Reservoirs

Abstract

15 water quality indicators of the deepwater layered Jinpen Reservoir were monitored from August 2011 to July 2012. The cluster analysis method was employed to explore the relationship between the outflowing water quality of the reservoir and the water quality of each longitudinal water layer to provide technical solutions for predicting the outflowing water quality of the reservoir and reducing the cost of water purification in the water plants effectively. The results indicate that by carrying out the K-means cluster analysis on the water quality of the longitudinal water strata according to the 3 cluster sets, the cluster centers of the cluster groups where the water strata of the withdrawal intake outlets are that are counted up can predict outflowing water quality of the water tower effectively; the degree of consistency of the predicted temperature, pH, ORP, conductivity, salinity, alkalinity with the actual test data is relatively higher, and the relative errors were within the-10% to 10%. Polynomial curve fitting was done for the upper and lower boundaries of the cluster group where the water strata of the withdrawal intake outlets are, and the outflow concentration obtained conformed to the theoretical derivation and the actual situation.