Event-Triggered Distributed Control of Nonlinear Interconnected Systems Using Online Reinforcement Learning With...

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Nonlinear Dynamics

Background:

• Interconnected nonlinear systems with uncertainties pose significant control challenges.
• Event-triggered control reduces communication load but requires sophisticated design.
• Approximate dynamic programming offers a framework for optimal control of complex systems.

Purpose of the Study:

• To develop a distributed control scheme for uncertain input affine nonlinear interconnected systems.
• To implement event-triggered state feedback using a hybrid learning scheme.
• To achieve near-optimal control policies with enhanced stability and reduced computational load.

Main Methods:

• Utilized approximate dynamic programming with a novel hybrid learning scheme.
• Employed artificial neural networks (NNs) for function approximation and system dynamics identification.
• Derived NN weight tuning rules and event-triggering conditions using Lyapunov stability theory.
• Introduced a novel NN weight update for approximating the optimal value function, considering NN approximation and bootstrapping effects.
• Incorporated an exploration strategy within the online control framework.

Main Results:

• Generated an approximate solution to the Hamilton-Jacobi-Bellman equation.
• Derived near-optimal control policies for individual subsystems.
• Achieved regulation of system states and NN weight estimation errors.
• Demonstrated local uniformly ultimately bounded stability.
• Validated analytical results through simulation studies.

Conclusions:

• The proposed hybrid learning scheme effectively controls uncertain nonlinear interconnected systems with event-triggered feedback.
• The approach ensures system stability and optimizes performance while managing computational resources.
• The integration of online exploration enhances the overall cost reduction during the learning phase.