A Study of Accuracy and Application of Genetic Programming for estimation of Scour Depth around Bridge Piers, below pipelines and Downstream of Bucket Overflow

Scour is a significant issue in the safety of hydraulic structures such as the dam’s downstream of bucket overflow, lower valve downstream, bridge piers and below pipelines. Many laboratories have studied the effect of different variables on the scour depth of these structures. Various models like neural networks (ANN), categorizing data (GMDH) and regression equations predicted the scour depth in sensitive hydraulic structures. In this study, the accuracy and application of genetic programming (GP) in the downstream of the dam’s overflow, bridge piers and below pipelines is examined. From these examinations the achieved result was that the GP model, when utilized to estimate the downstream scour depth of bucket overflow, has a higher coefficient of determination (= 0.977R^2) than when applied to the bridge piers and below pipelines. When this model is used for the estimation of the scour depth below pipelines, it has a lower absolute error (δ=9.9) than the other two cases. The root-mean-square error (RMSE) of GP model for prediction of the scour depth below pipelines is less than the other cases. In the bridge piers, GP model has the least root-mean-square error. When the GMDH model is taught with genetic programming, the outcome will have a higher coefficient of determination.

Scouring, GP Model, Bridge Piers, Overflow, Pipelines