scholarly journals Saturation-allowed zeroing neural networks activated by various functions for time-varying quadratic programming

Filomat ◽  
2020 ◽  
Vol 34 (15) ◽  
pp. 5149-5157
Author(s):  
Xiaoyu Guo ◽  
Mei Liu ◽  
Qinglin Zhao ◽  
Bin Hu ◽  
Huiyan Lu ◽  
...  
Keyword(s):  
Neural Networks ◽  
Quadratic Programming ◽  
Convex Function ◽  
Activation Function ◽  
Time Varying ◽  
Unbounded Function ◽  
Model Based ◽  
Saturation Function ◽  
Simulation Results ◽  
Increasing Functions

Zeroing neural networks (ZNN) approach, has been presented to solve a lot of time-varying problems activated by monotonically increasing functions. However, the existing ZNN models for timevarying quadratic programming based on ZNN approach may be different from each other in structures, but share two common restrictions, i.e., the function must be convex and unbounded. In order to relax the above restrictions in solving time-varying quadratic programming (TVQP) problems, this paper proposes a saturation-allowed zeroing neural networks (SAZNN) model based on the ZNN approach. Comparing with existing models, the activation function (AF) of SAZNN model tolerates more kinds of functions, e.g., saturation function, non-convex function and unbounded function. Finally, this paper provides simulation results synthesized by the proposed SAZNN model activated by various AFs and verifies the superiority of the proposed SAZNN model in terms of convergence, efficiency and stability.

Adaptive Synchronization in Unknown Stochastic Chaotic Neural Networks with Mixed Time-Varying Delays

2011 ◽  
pp. 289-313
Author(s):  
Jian-an Fang ◽  
Yang Tang
Keyword(s):  
Neural Networks ◽  
Sufficient Conditions ◽  
Stochastic Perturbation ◽  
Adaptive Synchronization ◽  
Output Coupling ◽  
Time Varying ◽  
Connection Weight ◽  
Chaotic Neural Networks ◽  
Simulation Results ◽  
Feedback Techniques

Neural networks (NNs) have been useful in many fields, such as pattern recognition, image processing etc. Recently, synchronization of chaotic neural networks (CNNs) has drawn increasing attention due to the high security of neural networks. In this chapter, the problem of synchronization and parameter identification for a class of chaotic neural networks with stochastic perturbation via state and output coupling, which involve both the discrete and distributed time-varying delays has been investigated. Using adaptive feedback techniques, several sufficient conditions have been derived to ensure the synchronization of stochastic chaotic neural networks. Moreover, all the connection weight matrices can be estimated while the lag synchronization and complete synchronization is achieved in mean square at the same time. The corresponding simulation results are given to show the effectiveness of the proposed method.

New criteria for dissipativity analysis of Caputo fractional-order neural networks with non-differentiable time-varying delays

2022 ◽  
Vol 0 (0) ◽  
Author(s):  
Nguyen Thi Phuong ◽  
Nguyen Thi Thanh Huyen ◽  
Nguyen Thi Huyen Thu ◽  
Nguyen Huu Sau ◽  
Mai Viet Thuan
Keyword(s):  
Neural Networks ◽  
Fractional Order ◽  
Linear Matrix ◽  
Sufficient Condition ◽  
Time Varying ◽  
Matrix Inequalities ◽  
Simulation Results ◽  
Delay Dependent ◽  
Dissipativity Analysis ◽  
New Criteria

Abstract In this article, we investigate the delay-dependent and order-dependent dissipativity analysis for a class of Caputo fractional-order neural networks (FONNs) subject to time-varying delays. By employing the Razumikhin fractional-order (RFO) approach combined with linear matrix inequalities (LMIs) techniques, a new sufficient condition is derived to guarantee that the considered fractional-order is strictly (Q, S, R) − γ − dissipativity. The condition is presented via LMIs and can be efficiently checked. Two numerical examples and simulation results are finally provided to express the effectiveness of the obtained results.

Time-Varying Delay Global Stability of Neural Networks

2011 ◽  
Vol 65 ◽  
pp. 9-12
Author(s):  
Dong Ming Huang ◽  
Lei Sun
Keyword(s):  
Neural Networks ◽  
Exponential Convergence ◽  
Activation Function ◽  
Time Varying ◽  
Functional Stability ◽  
Neural Network Analysis ◽  
Connection Weight ◽  
Time Varying Delay ◽  
Exponentially Stable ◽  
Varying Delay

Hopfield neural networks with variable delay stability of the equilibrium point, the delayed neural network analysis of exponential convergence rate and exponential stability. Obtained by using Lyapunov functional stability of the index to determine the conditions, the use of a number of analytical methods to study the connection weight matrix and activation function of the boundary, has been the result of system is exponentially stable, and a numerical example to prove that the method effectiveness.

Nonlinear Ethanol Gasoline Optimal Control System Based on Hammerstein Model

2013 ◽  
Vol 765-767 ◽  
pp. 1889-1892
Author(s):  
Zhong Ming Luo ◽  
Zhuo Fu Liu ◽  
Fan Wang ◽  
Ling Sen Lin
Keyword(s):  
Neural Networks ◽  
Optimal Control ◽  
Control System ◽  
Internal Models ◽  
Hammerstein Model ◽  
Optimal Control System ◽  
Model Based ◽  
Simulation Results ◽  
Feedback Neural Networks

In this paper, a Hammerstein model based on forward feedback neural networks was proposed to tackle the optimal control of a nonlinear MISO system. The method offers a solution to the optimization of internal models. The optimal control with the preset value was implemented under both static and dynamic optimal indices. The simulation results showed that the algorithm can fulfill the task of blending ethanol and gasoline effectively.

scholarly journals Superior performance of using hyperbolic sine activation functions in ZNN illustrated via time-varying matrix square roots finding

2012 ◽  
Vol 9 (4) ◽  
pp. 1603-1625 ◽  
Author(s):  
Yunong Zhang ◽  
Long Jin ◽  
Zhende Ke
Keyword(s):  
Neural Network ◽  
Computer Simulation ◽  
Activation Function ◽  
Superior Performance ◽  
Time Varying ◽  
Activation Functions ◽  
Square Roots ◽  
Hyperbolic Sine ◽  
Simulation Results ◽  
Large Model

A special class of recurrent neural network (RNN), termed Zhang neural network (ZNN) depicted in the implicit dynamics, has recently been proposed for online solution of time-varying matrix square roots. Such a ZNN model can be constructed by using monotonically-increasing odd activation functions to obtain the theoretical time-varying matrix square roots in an error-free manner. Different choices of activation function arrays may lead to different performance of the ZNN model. Generally speaking, ZNN model using hyperbolic sine activation functions may achieve better performance, as compared with those using other activation functions. In this paper, to pursue the superior convergence and robustness properties, hyperbolic sine activation functions are applied to the ZNN model for online solution of time-varying matrix square roots. Theoretical analysis and computer-simulation results further demonstrate the superior performance of the ZNN model using hyperbolic sine activation functions in the context of large model-implementation errors, in comparison with that using linear activation functions.

Sign in / Sign up

Export Citation Format

Share Document