Detection of Sediment Transport with Artificial Intelligence Models Formed by Using K-Means Clustering Algorithm

It isimportant to detect sediment load truly in the project process of water structures such as both dam and regulator to determine pollution level, to use and to supply potable water. In this work, methods such as Adaptive Neuro Fuzzy Interference Systems (ANFIS), Artificial Neural Networks (ANN) and Multiple Linear Regression (MLR) are experienced for three Flow Observation Stations on Fırat Basin. K-Means Clustering Algorithm is used to select Clustering numbers of ANFIS Models being formed. Sediment (Qs), Precipitation (P), Flow rate (Q) and temperature (P) data from each station are used in this study. Sediment estimation model is developed for each station by using the data. The value of the day when precipitation occurs (P), the value of the day before precipitation occurs (P-1), flow and temperature rates are used as input factors; sediment concentration is used as output factors. The model being formed is compared for all of the stations with regard to regression coefficient (R2 ) and mean percentage error (MPE) in both education and test stages. As a consequence of the analysis, subset numbers of ANFIS model being formed by determining subset of K-Means Clustering Algorithm has more successful results instead of ANFIS Models being randomly formed. It is also observed that the Models of ANFIS and ANN methods have closer results to the values being observed when they are compared to MLR methods.