Application of Neuro-fuzzy Method for Prediction of Vehicle Fuel Consumption

This paper presents the application of neuro-fuzzy method for prediction of vehicle fuel consumption prediction. Prediction motor vehicle fuel consumption has become a strategic issue, because it is not only related to the issue of availability of fuel but also the problem of the environmental impact caused. This study used automobile data, i.e. number of cylinders, displacement, horsepower, weight, acceleration, and model year, while the output variable to be predicted is the fuel consumption in MPG (miles per gallon). 'Weight' and 'Year' are selected as the best two input variables. The training and checking errors are getting distinguished, indicating the outset of overfitting. The results of this research are expressed in three dimension input-output surface graph of the best two-input ANFIS model for MPG prediction. It is a nonlinear and monotonic surface, in which the predicted MPG increases with the increase in 'Weight' and decrease in 'Year'. The training RMSE is 2.767; the checking RMSE is 2.996. The greater the weight of the motor vehicle, the greater the amount of fuel needed to travel the same distance. In comparison, a simple linear regression using all input candidates results in a training RMSE of 3.453, and a checking RMSE of 3.445.