Mixed-Gain Adaption-Based Fault-Tolerant Funnel Control of Robotic Manipulators With Unknown Dynamics and Sensor Faults

• Published in: IEEE transactions on automation science and engineering 2024, p.1-12
• Main Authors: Zhang, Jin-Xi, Song, Jun-Guo, Chen, Qingda, Yang, Feng
• Format: Article
• Language: English
• Subjects: Fault-tolerant control; predefined performance; reference tracking; robotic manipulators; sensor faults
• ISSN: 1545-5955; 1558-3783
• DOI: 10.1109/TASE.2024.3388592

This paper is dealt with the problem of reference tracking for the robotic manipulators with unknown dynamics and subject to multiplicative and additive sensor faults. The resulting dynamics of the closed-loop system is both unknown and structurally variable, for which the widely adopted approximation tools are not effective straightforward. Moreover, the multiplicative sensor faults result in the loss of strong controllability of the closed-loop system, which renders the conventional fault-tolerant control methods infeasible. To overcome these obstacles, an innovative fault-tolerant funnel control strategy based on a mixed-gain adaption technique is put forward in this paper. The controller involves information of neither the plant model nor the sensor faults, but does not invoke the tools for parameter identification, function approximation, data driving, iterative learning, disturbance estimation, fault estimation or fault diagnosis. It achieves reference tracking of the robotic arms with the predefined settling time and accuracy against the sensor faults. A comparative experiment on a 2-DoF serial flexible link robot is conducted to illustrate the effectiveness and advantages of the proposed approach. Note to Practitioners -This paper was motivated by the problem of fault-tolerant control (FTC) of robotic manipulators but also applies to other MIMO nonlinear plants with inputs coupling, e.g., permanent magnet synchronous motors and dynamic positioning operated vessels. The existing FTC methods work for the case where sensor faults do not alter the strong controllability of the closed-loop system. Instead, the non-strongly controllable faulty plants are considered in this paper to develop the FTC approaches to further enhance the fault tolerance of robotic manipulators. On the other hand, the involvement of sensor faults results in irregular dynamics of the closed-loop system, which cannot be addressed by the conventional approximation techniques, e.g., neural networks or fuzzy logic systems. Therefore, FTC of robotic manipulators against sensor faults that cause loss of strong controllability of control systems is challenging but significant in both academia and industry. To this end, this paper suggests a novel fault-tolerant funnel control approach based on a mixed-gain adaption technique. It is independent of the model information or the fault information, thus with a high universal property. Even so, no effort is paid for model identification or fa