Estimation of live bed scour depth around different shapes of bridge piers using ANFIS and SVMR approach

Abstract

The phenomenon of local scour from around the bridge pier is considered to be a compound process, because the safety of the bridge is proportional to the quantity of local scour depth occurred around the bridge pier. Hence, it is inconvenient to adopt a standard method to estimate and predict the accurate scour depth. In the present discussion, an effort is made using Artificial Intelligence (AI) models/tools, namely, Neuro-Adaptive-Inference-System (ANFIS) and Support Vector Machine Regression technique (SVMR). To understand thoroughly, the local scour taking place around the bridge pier using AI techniques the input parameters which influence the depth of the scour under the live bed condition are sediment quantity, flow velocity and time of flow. The ANFIS with Gauss membership function (MF) and SVM with Radial Basis Function (RBF) kernel function model predictions are analyzed using model performance indicators such as RMSE, CC, NSE, etc. The results are compared with the measured scour depths of different pier shapes such as Circular; Rectangular; Round nosed and Sharp-nosed pier. The simulated scour depth depicts that ANFIS with Gauss MF model is found to be reliable and efficient in predicting the scour depth around all four types of bridge pier shapes under live-bed condition.