Fuzzy Hybrid Neural Network Control for Uncertainty Nonlinear Systems Based on Enhancement Search Algorithm
Hiep-Binh Nguyen,Chih-Min Lin,Tuan-Tu Huynh,Hsing-Yueh Cho,Duc-Hung Pham,Fei Chao,Ha Le Nhu Ngoc Thanh
DOI: https://doi.org/10.1007/s40815-022-01374-0
IF: 4.085
2022-01-01
International Journal of Fuzzy Systems
Abstract:This study aims to propose a more efficient control system for uncertain nonlinear systems. A new fuzzy hybrid neural network named as backstepping self-organizing hybrid function-link fuzzy brain emotional learning controller (BSDFFBC) is developed. The proposed hybrid structure is a type of brain-imitated neural network. The BSDFFBC is combined with a backstepping control technique, an enhancement search algorithm (ESA), and a hybrid function-link network (HFLN) to enforce its control ability. Moreover, a self-organizing mechanism is used to automatically adjust the number of neurons that can maintain the BSDFFBC to accommodate large variations in inputs and system uncertainties and can also significantly reduce the computation time. Next, a new genetic algorithm-based ESA is proposed to find the optimal parameters for the control system. Then, the online learning rules of the BSDFFBC are designed using the backstepping control technique and the gradient descent algorithm. Thus, the system structure and control parameters of the control system can be online adjusted to achieve more efficient control performance. Moreover, a robust compensation controller is added to BSDFFBC to improve the control quality of the system. Finally, the proposed BSDFFBC control system is used to control a roll-to-roll system and a magnetic ball levitation system to show its favorable control performance; and the comparisons with other controllers have shown its superiority.