Engine performance and exhaust gas temperature modeling in a diesel engine with butanol-diesel fuel blends: Comparison of artificial neural networks and regression analysis

In this study, the effect of using a mixture of butanol and diesel fuel in a small diesel engine on engine performance and exhaust emissions is modeled. A comparison has been made with artificial neural networks, which is an up-to-date method, and multiple linear regression methods. Motor performance parameters, torque, effective power and brake specific fuel consumption are used as dependent variables. In addition, the exhaust gas temperature, which is an important parameter in the engines, is taken as another dependent variable. In this study, independent variables are selected as butanol-diesel fuel mixture ratio and engine speed. In order to compare two different modeling techniques, the mean squared error and mean absolute percent error are calculated. As a result of this study, artificial neural networks give better results than multiple linear regression techniques for four different dependent variables.