COMPARISON OF MEMBERSHIP FUNCTIONS IN ADAPTIVENETWORK- BASED FUZZY INFERENCE SYSTEM (ANFIS) FOR THE PREDICTION OF GROUNDWATER LEVEL OF A WATERSHED

Abstract

In this study, Adaptive-Network-Based Fuzzy Inference System (ANFIS) approach, introduced by Jang (1993) was employed to investigate its applicability in predicting water level in Thurinjapuram watershed, Tamilnadu, India. The ANFIS combines the explicit knowledge representation of Fuzzy Inference System with the learning power of Artificial Neural Networks, therefore it is a very powerful approach to build complex relationship between a set of input and output data. It also provides a natural framework for combining both numerical information in the form of input/output pairs and linguistic information in the form of IF–THEN rules in a uniform fashion. The results of the modeling were reasonable in terms of statistical performances. Several indices of performance such as mean average error, root mean squared error, and coefficient of correlation showed good performance. The results were compared with different type of membership function. The model with Gaussian membership functions gave the best performance among all given models. The capability ANFIS system in producing such reasonable result indicated that this approach has potentially to be used as predictor of water level in the Thurinjapuram watershed, Tamilnadu, India.