Utilizing the ANFIS and teaching-learning-based optimization (TLBO) methods to predict the maximum depth of scouring at culvert outlets

Abstract

ABSTRACT Excessive scouring at culvert outlets is a widespread issue that often leads to culvert failure. At present, there is an absence of a universally applicable model for accurately predicting the occurrence of local scouring at culvert outlets. In this paper, a comparison is made between the performance of existing empirical equations and the results obtained from two advanced methods: adaptive neuro-fuzzy inference system (ANFIS) and the teaching-learning-based optimization (TLBO). This paper presents the development of the TLBO method, based on the ANFIS approach, for simulating scouring at culvert outlets. The algorithms for supervised training are applied by utilizing data collected from published studies. The results illustrate that the ANFIS–TLBO model successfully predicts scouring depth at culvert outlets with higher accuracy in comparison to existing empirical formulas. Furthermore, the model exhibits a wider range of applicability, accommodating diverse conditions. By integrating the ANFIS approach and TLBO optimization, the model surpasses the limitations of traditional empirical equations, providing engineers with a more reliable tool for predicting scour depth at culvert outlets. Furthermore, the proposed model has the largest accuracy compared to the previously developed empirical models. Ultimately, it can be concluded that the hybrid ANFIS–TLBO is a robust approach for scour depth prediction downstream of culverts.