Spillways are one of the most common hydraulic structures used in water engineering projects. In this research, a group method of data handling (GMDH) was developed to estimate the amount of energy dissipation of the flow passing over the nonlinear weirs with triangular and trapezoidal plans. Next, to compare the performance of this model with other soft computing models, the artificial neural network model was developed as the most common soft computing model in the field of water engineering studies. Among the dimensionless parameters, the upstream relative head, the number of cycles, the Froude number, and the magnification ratio as input variables of models were used. The results showed that the statistical indicators of GMDH model error in the development stage are R2 = 0.913, RMSE = 0.010(training), and in the validation stage R2 = 0.829, RMSE = 0.015. The accuracy of the neural network model in the training stage is R2 = 0.957, RMSE = 0.007, and in the testing phase, R2 = 0.945, RMSE = 0.009. Examining the structure of the developed GMDH model shows that the relative upstream head parameter, the number of cycles, as well as the magnification ratio, play the most important role in the development of the network.