Adaptive terminal super-twisting prescribed performance controller for near-space vehicle based on data-driven model.

Area of Science:

• Aerospace Engineering
• Control Systems
• Artificial Intelligence

Background:

• Near-space vehicles (NSVs) require advanced control for stable flight and efficient scramjet operation.
• Traditional controllers struggle with model uncertainties and external disturbances, potentially leading to performance degradation and scramjet choking.

Purpose of the Study:

• To design a data-driven adaptive terminal super-twisting prescribed performance controller (DASTPC) for NSVs.
• To ensure transient and steady-state performance while preventing scramjet choking.
• To enhance control robustness against model uncertainties and external disturbances.

Main Methods:

• Proposed a novel predetermined-time performance function for faster, more accurate error convergence.
• Developed a non-singular fast terminal sliding surface and an adaptive super-twisting reaching law for improved efficiency and reduced chattering.
• Utilized a deep recurrent neural network (LSTM) for a data-driven model to handle NSV dynamics and uncertainties.
• Employed a homogeneous high-order sliding mode observer to compensate for external disturbances.

Main Results:

• The DASTPC guarantees tracking error convergence within a predetermined time.
• The adaptive reaching law effectively tunes control gain, mitigating chattering and excessive gains.
• The LSTM-based data-driven model successfully suppresses model uncertainties.
• The controller effectively restricts the angle of attack, ensuring scramjet intake conditions.

Conclusions:

• The DASTPC demonstrates superior performance in simulations compared to traditional methods.
• The controller achieves prescribed performance bounds under disturbances and parameter perturbations.
• The proposed approach enhances NSV flight safety and operational efficiency.