Adaptive Cruise System Based on Fuzzy MPC and Machine Learning State Observer.

Area of Science:

• Automotive Engineering
• Control Systems
• Artificial Intelligence

Background:

• Vehicle intelligentization necessitates advanced control systems for enhanced safety and comfort.
• Traditional Adaptive Cruise Control (ACC) systems exhibit limitations in tracking performance, comfort, and robustness within uncertain and dynamic driving environments.

Purpose of the Study:

• To propose a hierarchical control strategy to improve ACC system performance under uncertain conditions.
• To enhance tracking accuracy, passenger comfort, and control robustness of ACC systems.

Main Methods:

• A deep learning-based dynamic normal wheel load observer was integrated into the ACC perception layer.
• A Fuzzy Model Predictive Control (fuzzy-MPC) approach was employed for controller design, optimizing performance indicators and adapting to changing scenarios.
• An integral-separate PID controller was utilized for the executive layer, alongside a rule-based ABS control method for improved safety.

Main Results:

• The proposed hierarchical control strategy demonstrated superior tracking accuracy and stability compared to traditional methods in simulations.
• The dynamic normal wheel load observer effectively informed brake torque allocation.
• The fuzzy-MPC controller successfully adapted to varying driving conditions by adjusting objective function weights and constraints.

Conclusions:

• The developed hierarchical control strategy significantly enhances ACC system performance, offering improved safety, comfort, and robustness.
• The integration of deep learning and fuzzy-MPC provides a promising direction for future intelligent vehicle control systems.
• The strategy effectively addresses the limitations of conventional ACC systems in dynamic and uncertain driving scenarios.