No abstract

No abstract

No abstract

Scour depth prediction is an important aspect of designing a bridge pier structure in a river. Proper modeling of scour depth ensures the sustainability of the structure. An attempt is made to develop a scour depth model for the bridge pier using an adaptive network-based fuzzy inference system (ANFIS) and gene expression programming (GEP). The scour depth is found to be influenced by various independent parameters such as pier diameter, flow depth, approach mean velocity, critical velocity, Froude number, bed sediment, and geometric standard deviation of bed particle size. Gamma tests are performed to identify the best input parameter combinations to predict scour depth. In the present study, two separate models have been developed for clear-water scouring (CWS) and live-bed scouring (LBS). For different ranges of input parameters, the scour depth ratio is computed and error analysis is performed. Results indicate that the ANFIS model (R2CWS = 0.95, MAPECWS = 9.39% and R2LBS = 0.95, MAPELBS = 5.29%) is the most accurate predictive model in both scour conditions as compared to the GEP model and existing models of previous researchers. However, for the low value of pier diameter (b) to flow depth (y) ratio (<0.25), the present ANFIS model apportioned unsatisfactory results for LBS only.

The analytical methods require the system of non-linear equations to be solved which are very complex. So, mathematical models that prompt in taking care of complex system of problem are solved here through an artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS). By utilizing ANN and ANFIS, an attempt is taken to predict the discharge in converging and diverging compound channel. Gamma test and M test have been performed to achieve the best combinations of input parameters and training length respectively. The significant input parameters that influence the discharge are found to be friction factor ratio, hydraulic radius ratio, relative flow depth, and bed slope. A suitable performance is achieved by the ANFIS model as compared to ANN model with a high coefficient of determination of 0.86 and low root mean square error of 0.005 in predicting the discharge of non-prismatic compound channels taken under consideration.