Application Of Adaptive Neuro-fuzzy Inference System For The Estimation Of Roughness Coefficient Of A Meandering Open-channel Flow
Free (open access)
Volume 10 (2015), Issue 1
87 - 99
S. MOHARANA, K. K. KHATUA & M. SAHU
An experimental investigation concerning the variation in roughness for meandering channels with flow depths, aspect ratio and sinuosity is presented. Test results revealed that the value of roughness coefficient in terms of Chezy’s C increases with increase in aspect ratio and sinuosity. Adaptive neuro-fuzzy-based inference system (ANFIS), an integrated system, a combination of fuzzy logic and neural network is employed to find out the roughness coefficient of a meandering channel. Estimation of roughness coefficient is important for forecasting of discharge because its flexibility to resolve issues supported nonlinearity, randomness and uncertainty of knowledge. In the present work, an ANFIS-based model is developed for the prediction of the roughness coefficient of a meandering channel in terms of Chezy’s C. Different standard methods to predict this variable are conjointly tested and verified with the laboratory findings as well as global data moreover. By comparing the results with the established standard methods available in the literature, it was observed that traditional methods could not provide satisfactory output at different surface and hydraulic conditions. Statistical error analysis is also carried out in which it was found that ANFIS model performed more accurately giving results with less error than different existing strategies. The analysis shows a high level of accuracy with regard to the ANFIS-based model developed for predicting the Chezy’s C especially coefficients of determination are found to be more encouraging.
Aspect ratio, flow resistance, fuzzy inference system, meandering channel, sinuosity