Exponentially convergent state estimation for delayed switched recurrent neural networks.

Area of Science:

• Control Theory
• Artificial Neural Networks
• Systems Engineering

Background:

• Switched neural networks are crucial in complex systems but are susceptible to delays.
• Accurate state estimation is vital for performance and stability analysis in these networks.
• Existing methods often struggle with delay-dependent stability in switched systems.

Purpose of the Study:

• To develop a delay-dependent state estimation method for switched neural networks.
• To ensure exponential convergence and stability of the estimation error system.
• To provide a computationally tractable approach for designing state estimators.

Main Methods:

• Derivation of delay-dependent criteria for exponential stability.
• Formulation of the state estimator gain matrix using linear matrix inequalities (LMIs).
• Utilizing standard numerical packages for LMI solutions.

Main Results:

• Established delay-dependent criteria guaranteeing exponential stability of the estimation error system.
• The proposed state estimator's gain matrix is efficiently computable via LMIs.
• Demonstrated the effectiveness of the developed state estimator through an illustrative example.

Conclusions:

• The proposed method effectively addresses the delay-dependent state estimation problem in switched neural networks.
• The LMI-based approach offers a practical and computationally efficient solution.
• This work contributes to robust control and estimation for complex dynamical systems.