Finite-time parameter identification of fractional-order time-varying delay neural networks based on synchronization

Area of Science:

• Computational Neuroscience
• Control Theory
• Artificial Intelligence

Background:

• Fractional-order systems offer more sophisticated modeling capabilities than integer-order systems.
• Time-varying delays introduce significant complexity in analyzing and controlling neural networks.
• Parameter identification is crucial for understanding and applying neural network models.

Purpose of the Study:

• To develop a finite-time parameter identification method for fractional-order time-varying delay neural networks (FTVDNNs).
• To achieve synchronization between drive-response FTVDNNs while simultaneously identifying uncertain parameters.
• To provide theoretical analysis and simulation verification for the proposed method.

Main Methods:

• Utilizing the fractional-order Lyapunov stability theorem.
• Designing a synchronous controller based on feedback control principles.
• Developing parameter update rules for adaptive identification.
• Conducting theoretical analysis to determine stable time.

Main Results:

• Successful synchronization of drive-response FTVDNNs in finite time.
• Accurate identification of uncertain parameters within the FTVDNNs.
• Validation of the method through two simulation examples.
• Calculation of the finite stable time for the synchronization process.

Conclusions:

• The proposed synchronization-based method effectively identifies parameters of FTVDNNs in finite time.
• The approach guarantees synchronization and parameter identification simultaneously.
• The method is robust and validated through simulations, applicable to complex neural network dynamics.