An optimal control algorithm toward unknown constrained nonlinear systems based on the sequential sampling and...

Area of Science:

• Engineering
• Control Theory
• Computational Methods

Background:

• Optimal control theory application in engineering is hindered by high modeling costs and complexity.
• Existing methods often struggle with real-world constraints and require accurate plant models.

Purpose of the Study:

• To develop a model-free direct method for solving optimal control problems with general constraints.
• To bridge the gap between theoretical optimal control and practical engineering applications.
• To reduce the reliance on detailed mathematical models of controlled systems.

Main Methods:

• A novel model-free direct method utilizing sequential sampling and surrogate model updating.
• The algorithm iteratively refines a surrogate model using actual trajectory data.
• New sampling strategies and stopping criteria are introduced to ensure convergence and trajectory overlap.

Main Results:

• The method successfully solves model-free optimal control problems with general constraints.
• Demonstrated ability to achieve accurate, constrained solutions with significantly fewer sample data points.
• Ensured overlap between final actual and planned trajectories through novel stopping criteria.

Conclusions:

• The proposed model-free direct method offers a practical and efficient solution for complex optimal control tasks.
• This approach reduces the burden of model development, making advanced control techniques more accessible in engineering.
• The algorithm provides a robust framework for achieving high-accuracy constrained control with minimal data requirements.