Model-based control system design is a well-established method for advanced
engine control systems. These control systems maintain engine operation at
levels that meet stringent environmental regulations on vehicular emissions.
However, the models required for model-based design need to be accurate enough
for design and pre-calibration and fast enough for optimization and
implementation purposes. On the other hand, the variable valve timing (VVT)
technology significantly affects the dynamic performance of internal combustion
engines (ICEs). This study aims at developing a control-oriented extended
mean-value model (EMVM) of a gasoline engine, taking into account the effects of
VVT on the dynamic model. The developed model analyzes the engine performance
characteristics in transient and steady-state regimes. The engine model
incorporates four peripheral, nonlinear, dynamic subsystems: manifold, fuel
injection, wall-film adhesion, and evaporation processes. Moreover, lying at the
core of the developed model is a nonlinear, static, in-cylinder process (ICP)
model which simulates gas exchange and combustion processes based on the
cylinder boundary conditions. Based on the experimental data obtained from the
engine test setup, an artificial neural network (ANN) has been trained to
predict the ICPs as a single model. The ICP model was integrated into the
dynamic peripheral models to form the final EMVM. The results of the developed
model were compared to the engine experimental tests for two test scenarios:
half-throttle and full-throttle cases. It was observed that the developed model
could accurately simulate the engine speed, inlet air pressure, aspirated air
mass, and exhaust temperature. Moreover, the EMVM could successfully predict the
effects of VVT on the performance of ICEs.