Adaptive fuzzy fault‐tolerant control for non‐linear systems under actuator and sensor faults: the practical fixed‐time stability

Abstract

This study investigates the adaptive fault-tolerate control for strict-feedback non-linear system, in which actuator and sensor faults may happen simultaneously. Moreover, the additive and multiplicative faults are considered, which cover the bias, drift, loss of accuracy, and loss of effectiveness faults. Under the multiple sensor faults, a modified backstepping technique involving in the faulted states is proposed by utilising the fuzzy approximation to the unknown non-linear functions. Later, the adaptive fuzzy fault-tolerant control strategy is designed to ensure the practical fixed-time stability of the closed-loop system. Meanwhile, the convergence time is independent of the initial states of the system, and the control performance can be ensured for arbitrary bounded initial states in this study. Finally, the Buck converter circuit is employed to demonstrate the effectiveness of the proposed fault-tolerant control scheme.

Model free based finite time fault‐tolerant control of robot manipulators subject to disturbances and input saturation

“…Remark 1. The ultra-local model presented in (11) is of the canonical form for M 0 is positive diagonal matrix and invertible, and the controllability and observability of (11) was identical as (6).…”

“…The dynamics of the robot manipulators with n degree of freedoms (D.O.Fs) can be expressed as, M q + C(q, q) q + G + 𝝉 d = 𝝉 (6) where M ∈ R n×n is the inertial matrix, q = [ q 1 , … , q n ] T ∈ R n , q ∈ R n , and q ∈ R n represent the vector of joint angles, joint velocity, and joint acceleration, respectively, C(q, q) ∈ R n×1 is the matrix of Coriolis and Centripetal term, G is the gravity term, 𝝉 = [𝜏 1 , … , 𝜏 n ] T ∈ R n×1 is vector of the driving torque, and 𝝉 d ∈ R n×1 is the lumped disturbances, and is expressed as,…”

“…2 In the passive FTC, the fault is considered as an additional disturbance of the system, and the controller with high robustness is designed to cope with the lumped disturbance. Since the passive FTC do not need the feedback of the fault information, this method is more computational efficient and presents less time-delay than the active FTC, and many modern control algorithms have been proposed in passive FTC, such as neural network control, 3,4 backstepping control, 5 fuzzy control, 6 adaptive control, 7 and sliding mode control (SMC). [8][9][10][11] However, few studies on the FTC considered the problem of input saturation, in addition, they perceive the input saturation as the disturbance, and the adaptive mechanism 12 or the NN 3 were used to estimate the disturbance, but the real characteristics of the input saturation was not demonstrated and the finite time convergence cannot be achieved.…”

Event‐triggered adaptive neural control for stochastic nontriangular structure nonlinear systems with multisensor and actuator failures: The practical fixed‐time stability

“…Although the designed controller is conservative, it is easier to implement from the practical application. Besides, compared with References 12,37,38, only one adaptive parameter is designed to reduce computation.…”

“…In addition, the most existing significant results have been either only studied actuator faults, [31][32][33][34] or only sensor faults, 35,36 or sensor and actuator faults in deterministic systems. 37,38 To the best of our knowledge, the FTC of the sensor and actuator are rarely studied simultaneously in SNSS with ETC. Hence, this motivates our current research.…”