Adaptive Neuro Fuzzy Inference System based Water Quality Index for Godavari River (India)

Abstract

Water Quality Index (WQI)  is a comprehensive tool, which is used to assess the water quality status of the streams. In the present study water quality status of the Godavari River at Nashik has been assessed at fourteen water quality monitoring stations (WQMS) located at bathing ghats and bridges along the course of the River. For Assessment of the water quality, prevalent National Sanitation Foundation Water Quality Index (NSFWQI) values have been determined at these WQMS using pH ,DO, BOD,Turbidity and Total solids parameter values. It is observed that NSFWQI has its inherent limitations and fails to give the representative results of water quality in the region of higher water pollution. To overcome this limitation, Artificial Intelligence techniques like Fuzzy logic, Artificial Neural Network (ANN) and ANFIS are employed. ANFIS approach integrates the merits of Fuzzy logic and ANN approach .In view of this, Adaptive Neuro Fuzzy Inference System (ANFIS) has been employed for the present study, using Triangular, Trapezoidal, Gaussian and Bell membership functions(MF). It has been observed that Gaussian and triangular MF gives better results for Eleven and three number of sampling stations, respectively. Therefore, the Gaussian MF model outperforms other MF models of the ANFIS approach for the study.