Parametric Neural Network-Based Model Free Adaptive Tracking Control Method and Its Application to AFS/DYC System

Area of Science:

• Control Engineering
• Artificial Intelligence
• Nonlinear System Dynamics

Background:

• Model-free nonlinear systems present significant challenges for accurate identification and control.
• Existing methods often struggle with unknown dynamics and modeling uncertainties.
• Trajectory tracking in complex systems requires robust adaptive control strategies.

Purpose of the Study:

• To develop an effective parametric neural network (PNN) for identifying unknown dynamics in model-free nonlinear systems.
• To design an adaptive tracking controller that compensates for identified nonlinearities and modeling errors.
• To ensure the stability and convergence of the closed-loop system using Lyapunov stability theory.

Main Methods:

• A novel parametric neural network (PNN) identifier with a parameter error-driven updating law for accurate and rapid system identification.
• An adaptive tracking controller combining feedback control for nonlinearity compensation and sliding mode control for modeling error management.
• Lyapunov stability analysis to guarantee the convergence of the integrated PNN identifier and adaptive controller.

Main Results:

• The proposed PNN identifier demonstrates high accuracy and speed in capturing unknown system dynamics.
• The adaptive tracking controller effectively compensates for system nonlinearities and uncertainties.
• Simulation results on an Automotive Front-End/Direct Yaw Control (AFS/DYC) system validate the approach's effectiveness.

Conclusions:

• The developed PNN-based adaptive control strategy offers a robust solution for trajectory tracking in model-free nonlinear systems.
• The parameter error-driven updating law enhances identification performance.
• The combined control approach ensures system stability and reliable tracking capabilities.