Artificial Neural Network-based emission control for future ICE concepts
The internal combustion engine contains several actuators to control the engine performance and emissions. These are controlled within the engine control unit and follow a certain operation strategy to achieve targets such as reduction of NOx emissions and fuel consumption. However, these two targets are contradictory, and a compromise is required. The operation state is depending on the system boundaries, such as engine speed, load, temperature levels and exhaust aftertreatment system efficiency. This leads to constantly changing target values to remain within the defined boundaries, in particular the legal emission limits. The conventional approach is using several operating modes. Each mode represents a specific compromise and is activated accordingly. To fulfil the emissions legislation, multiple modes are required, which increases the calibration efforts. This new control approach uses an artificial neural network that replaces the conventional multiple operation mode approach.