Advanced deep neural network-back-stepping controller to cancel the chaotic behavior of Duffing-Holmes oscillator

Read full text Summarise with VIRAT
Authors: WeiFang Liang, Ahmed Kateb Jumaah Al-Nussairi, Narinderjit Singh Sawaran Singh, Abdulqadir Ismail Abdullah, Monireh Ahmadi, Zhichen Wang
Cited by: 0
References: 36

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Chaotic systems are complex, constantly evolving systems that display unpredictable behavior due to their intrinsic randomness and sensitivity to initial conditions. They are characterized by the butterfly effect, where minor changes in starting conditions can lead to significantly different outcomes. Analysis of nonlinear behavior and chaotic control of the Duffing-Holmes system is investigated in this research and an optimal intelligent backstepping controller is designed to eliminate the chaotic movement outside normal behavior. The backstepping technique involves parameters that must be positive. Incorrectly choosing these parameters causes irrational behavior or system instability. In this research, a Deep Neural Network (DNN) is merged with the Backstepping method (DNN + Backstepping) To identify the suitable and optimal values for the controller parameters, one of the key benefits of the proposed control system is its ability to rapidly control chaos within a short time using a limited control signal. Numerical results demonstrate the effectiveness of this method in eliminating chaos from the Duffing-Holmes system. The results demonstrate a decrease of about 1 s in convergence time and an approximately 85% reduction in control signal amplitude. Moreover, the final convergence value of the objective function using the proposed method is significantly lower than unified particle swarm optimization (UPSO). In both of the methods, the state variables are converged to zero. By optimally tuning backstepping parameters via DNN, the method achieves state convergence to zero in 3 s—1 s faster than UPSO-Backstepping (4 s)—while reducing control signal amplitude by ∼85% (from peaks of ±0.6 to ±0.03) and minimizing the objective function by 99% (0.0002 vs. 0.31).

Fields of Study & Subjects

Chaos control and synchronization Adaptive Control of Nonlinear Systems Control and Stability of Dynamical Systems

Publication details

Journal / source
Energy Reports
Volume
15
Pages
109197–109197
Publication date
25 March 2026
Language
EN