Dynamic event-triggered H_∞ state estimation for discrete-time complex-valued memristive neural networks with mixed...

Area of Science:

• Control Theory
• Neural Networks
• Nonlinear Systems

Background:

• Memristive neural networks (MNNs) are crucial for advanced computing.
• Complex-valued MNNs (CVMNNs) offer enhanced capabilities.
• State estimation in CVMNNs with delays is challenging.

Purpose of the Study:

• To develop an H-infinity state estimator for discrete-time CVMNNs.
• To incorporate distributed and time-varying delays for practical modeling.
• To introduce a dynamic event-triggered scheme for efficient communication.

Main Methods:

• Conversion of CVMNNs to an augmented real and imaginary system.
• Design of a dynamic event-triggered state estimator.
• Lyapunov functional to guarantee estimation error system stability.

Main Results:

• A sufficient condition for asymptotical stability of the estimation error is derived.
• Explicit expression for the state estimator is obtained via matrix inequalities.
• The proposed method is validated through a simulation example.

Conclusions:

• The novel dynamic event-triggered H-infinity state estimation is effective for CVMNNs.
• The approach successfully handles delays and reduces communication burden.
• The method ensures the stability of the state estimation error.