Estimation of Runoff Through BPNN and SVM in Agalpur Watershed

摘要

Two different techniques support vector machine (SVM) and back propagation neural network (BPNN) employed to evaluate runoff for five proposed modeling inputs. Research is conducted at Agalpur watershed, Odisha, India. NSE, RMSE, and WI indicators are used for evaluation of performance of the model. Productivity of this work will propose development, plan, and administration of water-bound structures for mounting watershed. Presentation of this examiner is contrasted, and mapping is done with WI value. In BPNN, three different transfer functions like Tansig, Logsig, and Purelin are used to examine the model. Outcomes suggest that assessment of runoff is suitable to SVM in comparison to BPNN. Both BPNN and SVM perform well in complex data sets of projected watershed.