Modeling of the air conditions effects on the power and fuel consumption of the SI engine using neural networks and regression

Engine performance varies significantly due to the variations in weather conditions in different regions. So, in order to optimize the performance and fuel consumption, engines should be calibrated according to the weather conditions in which they operate. In this paper the effects of the air conditions (such as pressure and temperature) on the power and fuel consumption of the SI engine are modeled. First a comprehensive one-dimensional model of the real engine is constructed in GT POWER (R), and validated with experimental data from actual engine. Next, using this model, a set of experiments is carried out by varying pressure, temperature, and humidity of the incoming air, and engine speed. The measuring outputs are the power and BSFC of the engine. Then, two mathematical models are developed using MLP Neural Networks and also regression technique to estimate the outputs in terms of the inputs. At last, the estimation ability of the models is shown by a set of new experiments. These models could be used in engine calibration and shift the process from a near blind one to the one in which prior information have a significant role.