Data-driven control of hydraulic servo actuator: An event-triggered adaptive dynamic programming approach

Area of Science:

• Robotics and Control Systems
• Mechanical Engineering
• Artificial Intelligence

Background:

• Hydraulic servo actuators (HSAs) are critical for high-power, high-accuracy dynamic motion tasks in industry.
• HSAs are complex nonlinear systems with parameter uncertainties and unmeasurable states, posing control challenges.
• Accurate system identification is often difficult due to disturbances and parameter variations.

Purpose of the Study:

• To develop an event-triggered learning control strategy for HSAs with unknown dynamics using output feedback.
• To design a data-driven controller that does not require prior knowledge of system or exosystem dynamics.
• To improve control efficiency by reducing communication load and control updates via an event-based feedback mechanism.

Main Methods:

• Utilized adaptive dynamic programming (ADP) for online learning and control.
• Employed a linear discrete model of the HSA for controller design.
• Implemented a data-driven controller relying solely on measured input and output data.
• Introduced an event-based feedback strategy to optimize communication.

Main Results:

• The proposed ADP-based controller effectively handles HSAs with unknown dynamics.
• The data-driven approach eliminates the need for detailed system parameter knowledge.
• The event-triggered strategy significantly reduces communication resource usage and control updates.
• Theoretical convergence of the ADP algorithm was demonstrated.

Conclusions:

• The event-triggered adaptive dynamic programming approach offers a feasible and effective solution for optimal control of HSAs.
• This method provides robust control performance despite system uncertainties and unknown dynamics.
• The approach enhances practical applicability by leveraging readily available input-output data and reducing communication overhead.