Real time simulation of groundwater quality index using adaptive neuro-fuzzy inference

다국어 초록 (Multilingual Abstract)

o ensure sufcient quality and quantity of water for drinking, it is imperative to determine the contamination and quantifcation of potential damage of existing groundwater resources. The indexing of various water quality parameters and prediction of groundwater quality provide extensive technical assistance for the strategic management of groundwater resource. In this study, simulation of the groundwater quality index is carried out using adaptive neuro fuzzy inference system (ANFIS). The diferent combinations of input parameters are selected to develop the optimal model using grid partitioning and subtractive clustering FIS type. The architecture of ANFIS model is designed using Gaussian type membership function optimized through hybrid of back propagation and least square method. A total of 893 groundwater samples from 68 locations used for model development. The performance of models is weighed using correlation coefcient (R) and root mean square error.
The Model 4 consisting physio0chemical and anions produces R as 0.921 and 0.837 for training and testing. The results suggested that ANFIS is a robust model that could be used with high accuracy for the prediction of groundwater quality index.
Selection of adequate input parameter and ANFIS structure is a right approach for the prediction of groundwater quality and useful for decision makers for allocating water resources.