Design of a data-driven predictive controller for start-up process of AMT vehicles

Area of Science:

• Automotive Engineering
• Control Systems Engineering
• Data-Driven Modeling

Background:

• Automated Manual Transmissions (AMTs) present control challenges during vehicle start-up.
• Balancing efficiency (friction loss) and comfort (driveline shock) is critical for AMT performance.
• Existing control strategies may struggle with real-world uncertainties like varying vehicle mass.

Purpose of the Study:

• To design a data-driven predictive controller for AMT vehicle start-up.
• To achieve offset-free control and explicitly handle input/output constraints.
• To optimize the objective function for reduced friction and driveline shock.

Main Methods:

• A driveline simulation model was built using AMESim.
• Input-output data from the simulation model generated the controller.
• The predictor equation was derived using incremental inputs and outputs.
• Controller performance was evaluated under nominal and varied conditions.

Main Results:

• The proposed controller demonstrated effective performance during AMT start-up simulations.
• Achieved objectives included fast clutch lockup, reduced friction losses, and smooth vehicle acceleration.
• The closed-loop system exhibited robustness against uncertainties such as vehicle mass and road grade.

Conclusions:

• The data-driven predictive controller successfully manages the complexities of AMT vehicle start-up.
• The controller balances competing demands for efficiency and driver comfort.
• The system shows resilience to external disturbances, enhancing practical applicability.