Neuro-fuzzy modelling and forecasting in water resources

Abstract

Recently, with rapid development in computer science and technology, artificial intelligent (AI) models that emulate human thinking ability and brain structure are increasingly used in hydrological forecasting context. Neuro-fuzzy (N-F) models or specifically adaptive neuro-fuzzy inference systems (ANFIS) are rapidly becoming conventional in either academic or industrial applications. Although, there is a common network structure among ANFIS models, there is no one-fits-all ANFIS architecture for every case. Moreover, it is discussed that in many application, theory does not guide in model building process by either suggesting the relevant model input variables or correct functional form and model configuration. This paper is focused on the application of ANFIS in water resources context and reviews the common architecture of ANFIS models been used in this area of research. The aim is to familiarize the new researchers with ANFIS application process in water resources studies. Key words: Neuro-fuzzy modelling, adaptive neuro-fuzzy inference systems (ANFIS), hybrid learning algorithm, subtractive clustering.