Prediction of suspended sediment distributions using data mining algorithms

Abstract

Distribution of sediment concentration in open-channel flows, particularly in rivers, is one of the most important factors in understanding the river behavior, water quality, and design of hydraulic structures. Therefore, to determine the amount of transported suspended sediment, the sediment concentration distribution must be measured with high accuracy. In the present study, four intelligent methods of ANFIS-PSO, ANFIS-GA, ANFIS, and GMDH were used to predict the sediment concentration distribution. Since both GA and PSO optimization methods were used to optimize the ANFIS model, the performance of these models was significantly improved and their accuracies were increased. The results showed that the methods of ANFIS-PSO, ANFIS-GA, ANFIS, and GMDH were, respectively, the most accurate methods for prediction of suspended sediment distribution. Based on the evaluation of these methods, it was concluded that intelligent methods have considerable accuracy in predicting parameters affecting the suspended sediment distribution. Accordingly, considering the performance of these methods, a combination of optimization and intelligent methods may be useful for predicting sediment concentration distribution. It was also found that the ANFIS-PSO method can be a more appropriate and accurate method than other methods.

Keywords

• Sediment concentration distribution
• Intelligent methods
• Open channel flows