Improved adaptive neural network motion control for an aero-engine hydraulic system.

Area of Science:

• Aerospace Engineering
• Control Systems
• Artificial Intelligence

Background:

• Aero-engine hydraulic systems face uncertainties like complex aerodynamic forces and resource limitations.
• These disturbances impact system safety and performance.
• Existing control methods may struggle with limited bandwidth and computational power.

Purpose of the Study:

• To develop a novel control scheme for aero-engine hydraulic actuators.
• To address unknown disturbances and resource constraints effectively.
• To improve system performance while minimizing bandwidth occupation.

Main Methods:

• Utilized a radial-basis-function (RBF) neural network (NN) combined with an event-triggered technique.
• Implemented an event-triggered communication protocol for NN weight updates.
• Integrated a new motion control scheme with the designed RBF-NN estimator.

Main Results:

• The proposed controller ensures all closed-loop signals are ultimately bounded.
• Accurate trajectory tracking of the system output is achieved.
• The event-triggered approach prevents Zeno behavior and optimizes resource usage.

Conclusions:

• The combined RBF-NN and event-triggered control scheme is effective for aero-engine hydraulic systems.
• The method enhances estimation accuracy without excessive computational cost.
• Experimental validation on a Hardware-in-the-Loop platform confirms the control scheme's validity and benefits.