Resilient Asynchronous State Estimation for Markovian Jump Neural Networks Subject to Stochastic Nonlinearities and...

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Networked Systems

Background:

• Neural networks are crucial in complex systems, but their state estimation faces challenges from asynchronous data and uncertainties.
• Markov jump systems and networked-induced phenomena introduce complexities in real-time state estimation.
• Existing methods often struggle with simultaneous nonlinearities, sensor saturations, and parameter uncertainties.

Purpose of the Study:

• To develop a robust dissipativity-based asynchronous state estimation method for discrete-time Markov jump neural networks.
• To address randomly occurring nonlinearities, sensor saturations, and stochastic parameter uncertainties.
• To ensure the stability and performance of the estimator under realistic network conditions.

Main Methods:

• Modeling system nonlinearities using Bernoulli processes and statistical means.
• Employing a hidden Markov model to represent network communication environments and induced uncertainties.
• Developing a new criterion for stochastic stability and predefined dissipativity performance.
• Utilizing Lyapunov stability theory and linear matrix inequality (LMI) techniques.

Main Results:

• A novel asynchronous state estimation approach is proposed for the targeted neural network class.
• The developed criterion guarantees stochastic stability of the error system.
• Predefined dissipativity performance is achieved, ensuring system robustness.
• Simulation results validate the effectiveness of the proposed estimation method.

Conclusions:

• The proposed dissipativity-based asynchronous state estimation method effectively handles complex uncertainties in discrete-time Markov jump neural networks.
• The study provides a valuable theoretical framework and practical tool for robust state estimation in networked control systems.
• The findings contribute to the advancement of reliable state estimation techniques for intelligent systems operating in uncertain environments.