Adaptive Dynamic Surface Control of Epileptor Model Based on Nonlinear Luenberger State Observer

Area of Science:

• Neurology
• Computational Neuroscience
• Control Systems Engineering

Background:

• Epilepsy is a neurological disorder marked by recurrent seizures, often studied using computational models like the Epileptor.
• The Epileptor model simulates seizure dynamics but presents challenges due to its nonlinear, non-strictly feedback nature and inherent uncertainties.
• Accurate state estimation is crucial for controlling and understanding the Epileptor model, especially when only Local Field Potentials (LFPs) are measurable.

Purpose of the Study:

• To develop and validate an adaptive dynamic surface controller for the Epileptor model.
• To design a nonlinear Luenberger state observer for estimating unmeasurable states in the Epileptor model.
• To integrate Radial Basis Neural Networks (RBNNs) for nonlinear dynamics estimation within the observer-controller framework.

Main Methods:

• An adaptive dynamic surface controller was designed for the nonlinear Epileptor model.
• A nonlinear Luenberger state observer, utilizing RBNNs for nonlinear dynamics approximation, was developed to estimate system states from LFP signals.
• The stability of the closed-loop system (controller and observer) was rigorously proven using mathematical analysis.
• Performance was evaluated through simulations, demonstrating state and output tracking capabilities.

Main Results:

• The proposed adaptive dynamic surface controller and Luenberger state observer successfully estimated the Epileptor model's states.
• The integrated system demonstrated effective tracking of reference values for both states and outputs with acceptable error margins.
• Simulation results confirmed the stability and performance of the novel control and observation strategy.

Conclusions:

• The developed adaptive dynamic surface controller and Luenberger state observer represent a significant advancement in controlling and simulating the Epileptor model.
• This approach offers a robust method for state estimation and system control in computational epilepsy research, utilizing RBNNs for enhanced accuracy.
• The findings pave the way for more sophisticated modeling and potential therapeutic interventions for epilepsy based on computational dynamics.