Developing of ANFIS-Genetic Algorithm Meta-Heuristic Model for Predicting the Scour Depth in Vicinity of Submarine Pipelines

In coastal areas, passing oil and gas submarine pipelines is quite common and scouring around them threatens the stability of the submarine pipes. In this study, a meta-heuristic model is developed in order to predict the scour pattern in vicinity of the submarine pipelines. The model is produced using combination of adaptive Neuro-fuzzy inference system (ANFIS) and genetic algorithm (GA). Additionally, in this article, Monte Carlo simulations (MCs) were utilized to evaluate the accuracy of numerical models. On the other hand, in order to validate the numerical results, the k-fold cross-validation (k=6) was used. Next, six different numerical models were developed. Finally, by analyzing the numerical results, the superior model was introduced. The superior model simulated the scour depth with reasonable accuracy. The model simulated the scour depth by employing all input parameters. For example, correlation coefficient and scatter index for superior model were respectively calculated 0.974 and 0.090. In addition, distance between pipe and bed before scouring to pipe diameter (e/D) was identified as the most effective input parameter.