Swarm Intelligence-Based Support Vector Machine (PSO-SVM) Approach in the Prediction of Scour Depth Around the Bridge Pier

摘要

The mechanism of scour around the bridge pier is a complex phenomenon, and it is very difficult to make a common method to predict or estimate the depth of scour hole. In this paper, a hybrid model is developed, combining support vector machine and particle swarm optimization (PSO-SVM) to predict scour depth around a bridge pier. The input parameters such as sediment size (d_(50)), the velocity of flow (U), and time (t) are used in the study to predict the scour depth. The models are developed with RBF, polynomial, and linear kernel functions, and the performances are evaluated using different statistical parameters such as CC, RMSE, NSE, and NMB. The predicted results are compared with measured scour depth. The predicted scour depth reveals that PSO-SVM with RBF kernel function model is found to be reliable and efficient in predicting the scour depth around bridge piers.