Quasisynchronization for Neural Networks With Partial Constrained State Information via Intermittent Control Approach

Area of Science:

• * Neural Networks
• * Control Theory
• * Information Theory

Background:

• * Master-slave neural networks (MS NNs) often face challenges with mismatched parameters.
• * Limited communication channel (CC) capacity and controller efficiency are critical issues in network synchronization.

Purpose of the Study:

• * To achieve quasisynchronization (QS) in MS NNs with mismatched parameters under constrained communication.
• * To develop an efficient intermittent control strategy to manage CC capacity and controller performance.

Main Methods:

• * Utilized a logarithmic quantizer and round-robin protocol (RRP) to handle limited CC capacity.
• * Designed a transmission-dependent controller and established the synchronization error system (SES).
• * Developed a sufficient condition to ensure bounded QS for the MS NNs.

Main Results:

• * Successfully ensured bounded quasisynchronization for MS NNs with mismatched parameters.
• * The intermittent control strategy improved the efficiency of both the communication channel and the controller.
• * A concrete controller design method was provided.

Conclusions:

• * The proposed method effectively achieves QS in MS NNs under communication constraints.
• * The intermittent control strategy offers a viable solution for improving synchronization efficiency.
• * Numerical simulations validated the effectiveness of the developed approach.