Modeling and prediction of dilute ɣ-AL2O3/H2O nanofluids heat transfer coefficient using neuro-fuzzy system

In this research, nanofluids heat transfer in a circular tube at turbulent regime were simulated and predicted by neural and neuro-fuzzy networks. Normalized Re numbers and volume fractions were inputs data and normalized heat transfer coefficient was output data. The average relative error and mean square error have been calculated. These values in the neural network are 0.002 and 0.0005, respectively. And for training data of neuro-fuzzy method were 0 and 0, for test data were -0.0027 and 0.00067, respectively. Regression value in neural network was 0.99. This value in neuro-fuzzy method for training data was 1 and testing data was 0.988. According to regression values, neuro-fuzzy method is better than neural network method.