Adaptive Fuzzy Neural Tracking Control Of Euler-Lagrange Systems

Euler-Lagrange system is a representive system which variety of complicated dynamic systems can be well formulated by Euler-Lagrange systems.Research on tracking control of Euler-Lagrange systems has great theoretical value and practical significance.This paper presents three different adaptive fuzzy neural control strategies,which provide effective solutions for tracking control of Euler-Lagrange systems.First,a robust adaptive universe-based fuzzy backstepping tracking control is proposed,taking unmodeled uncertainties and external disturbances into account.Adaptive universe-based fuzzy systems are composed of general fuzzy basis functions with retractable fuzzy partitioning.By means of retractable fuzzy partitioning,the fuzzy partition of input space and fuzzy basis function can be reproduced automatically to adapt to the modified universe of discourse.Therefore,the tracking accuracy can be improved without adding fuzzy rules.Simulation studies demonstrate the effectiveness of the proposed scheme.Second,in order to reduce the dimension of the neural network inputs and lower computational complexity,a hybrid feedforward-feedback extreme learning robust adaptive tracking control scheme is proposed.Complicated system uncertainties can be effectively approximated by virtue of the established feedforward extreme learning neural network approximator.Compared with traditional feedback approximation approach,only desired trajectory is need as neural network input,which leads to a simpler approximator structure and reduces computational complexity significantly.Moreover,an H? robust term is further designed to eliminate the influence of unknown external disturbances and approximation errors.Simulation studies demonstrate the effectness of the proposed scheme.Finally,a self-constructing adaptive fuzzy neural network observer H? output feedback tracking control scheme for a class of Euler-Lagrange systems subject to complicated unmodeled dynamics and unknown external disturbances is proposed.It is considered that the position is available but the velocity is unavailable of the control system.A velocity observer based on self-constructing fuzzy neural network is designed to realize the exact estimation of the unavailable velocity.Particularly,fuzzy rules of the approximator can be generated online and redundant rules can be pruned,and thereby reducing computational complexity significantly.A sliding surface is further defined by combining tracking errors and its first derivatives,and system uncertainties and external disturbances are explicitly considered and merged into a simple lumped nonlinear term that can be accurately approximated by virtue of the proposed self-constructing adaptive fuzzy neural network approximator.Moreover,an H? robust compensation term is ulteriorly developed to effectively increase more accurate control performance and more robustness pertaining to the proposed control scheme.Simulation studies demonstrate the feasibility of the proposed scheme.