Mean Value and Real Time Engine Modeling

Contact name Mr. Miles Melka
Typical duration 1 day
Email for more information [email protected]

This course introduces mean value cylinder engine modeling and focuses on the method by which a detailed engine model can be simplified to a mean value engine model.  A mean value model uses a simplified neural network-based engine cylinder, as well as a simplified intake and exhaust system.  This results in a model that executes much more rapidly than the typical detailed GT-POWER engine model.  The speed of a mean value model simulation makes these types of models useful for vehicle or control system level transient simulations where computation speed is critical.  Mean value models may be used for both software in the loop (SiL) simulations or hardware in the loop (HiL) simulations.  The course will include introductions to Design of Experiments (DOE) and neural network control capabilities in GT-SUITE, as these tools are integral to mean value modeling.  Each student will participate in the class by converting an example detailed model to a mean value model.

  • Introduction to mean value modeling
  • Conversion of detailed model to mean value model
    • Characterizing the detailed model using DOE
    • Training neural networks to control mean value cylinder
    • Simplifying intake and exhaust systems
    • Model calibration and other considerations
  • Running a mean value model on an RT system
    • Preparing a mean value model to run RT
    • Setting up a coupled Simulink RT model
    • General description HiL capabilities and supported vendors

Note:  This course may be offered as a future open registration event OR requested as a custom training at a GTI office location or your location.