Application of adaptive Neuro-fuzzy inference system to estimate alongshore sediment transport rate (A real case study: southern shorelines of Caspian Sea)

Abstract

Many empirical models have been introduced by scientists during the recent decades for estimating longshore sediment transport rate, but these approaches have been calibrated and applied under limited conditions of the bed profile and specific range of the bed sediment size. The existing empirical relations are linear or exponential regressions based on the observation and measurements data and there's a great potential to build more accurate models to predict sediment transport phenomena by means of soft computation approach. This paper presents a novel case study application of the adaptive Neuro-fuzzy inference system (ANFIS) as a superior modeling technique for estimation of the longshore sediment transport rate in the southern shorelines of the Caspian Sea. The results will be compared with top three popular existing empirical equations. Daily grab samples from four stations were collected in the period of March 2012 through June 2012. The trained ANFIS model outperformed the existing regression-type empirical equations for the estimation of the alongshore sediment transport rate due to the adaptive structure of the ANFIS model to better fit complex systems.