Wu Yang,Yan-Wu Wang,Irinel-Constantin Morǎrescu,Xiao-Kang Liu,Yuehua Huang,Irinel-Constantin Morarescu
Abstract:In this brief, we investigate the fixed-time synchronization of competitive neural networks with multiple time scales. These neural networks play an important role in visual processing, pattern recognition, neural computing, and so on. Our main contribution is the design of a novel synchronizing controller, which does not depend on the ratio between the fast and slow time scales. This feature makes the controller easy to implement since it is designed through well-posed algebraic conditions (i.e., even when the ratio between the time scales goes to 0, the controller gain is well defined and does not go to infinity). Last but not least, the closed-loop dynamics is characterized by a high convergence speed with a settling time which is upper bounded, and the bound is independent of the initial conditions. A numerical simulation illustrates our results and emphasizes their effectiveness.
computer science, artificial intelligence, theory & methods,engineering, electrical & electronic, hardware & architecture
What problem does this paper attempt to address?
This paper aims to solve the fixed - time synchronization problem of competitive neural networks with multiple time scales (MTSCNNs). Specifically, the paper focuses on designing a new synchronization controller, which does not depend on the ratio between fast and slow time scales, making it easy to implement, and when the time - scale ratio is close to 0, the controller gain can still be well - defined and will not tend to infinity. In addition, this controller can make the closed - loop dynamics have a high convergence speed, and its settling time is bounded above, and this upper bound is independent of the initial conditions.
The main contributions of the paper are as follows:
1. A new type of synchronization controller is designed, which can achieve the synchronization of MTSCNNs within a fixed time, and its parameter settings are independent of the ratio of fast and slow time scales.
2. By setting appropriate controller parameters, the global fixed - time synchronization of the system is ensured, and at the same time, an upper - bound estimate of the settling time is provided.
3. The proposed method is not only suitable for theoretical analysis but also has practical application value, and the effectiveness of the method is verified by numerical simulation.
In the paper, the Lyapunov function method is used to prove the effectiveness of the designed controller, and specific numerical experiments are carried out to show that this method can achieve fixed - time synchronization under different initial conditions, further verifying the correctness and practicality of the theoretical results.