Low-Complexity Decentralized Output-Feedback Fault-Tolerant Control of General Unknown Interconnected Nonlinear Systems

This paper is concentrated on the problem of decentralized output-feedback control of interconnected strict-feedback systems with actuator failures. It is focused on the cases where the virtual control coefficients of the plant are unknown; the global boundedness, matching conditions or global Lipschitz conditions of the interconnections are not assumed; the control algorithm is as simple as possible. They render the existing decentralized output-feedback fault-tolerant control designs infeasible. To address the problem, a low-complexity decentralized robust prescribed performance control approach based on a linear state transformation and an input-driven filter is put forward in this paper. It achieves the system outputs to track the corresponding references with the preassigned speed and accuracy. It is also inherently robust against the unknown system dynamics, the actuator failures, and the disturbances, thus without parameter estimation, function approximation, derivative computation, command filtering, fault detection, fault isolation or fault estimation. Finally, a comparative simulation on two inverted pendulums linked by a spring is conducted to demonstrate the developed control design.
Note to Practitioners
—Many complex systems, such as power systems, aerospace systems, and chemical systems, can be modeled as interconnected systems. Moreover, due to the increasing scale and complexity of engineering systems, actuator failures are becoming more likely to occur during system operation. On the other hand, both the transient and steady-state tracking performance of the systems are required to be preassigned in practical scenarios, e.g., missile interception. Existing approaches to compensate for the actuator failures guarantee only the boundedness of the tracking error under nonparametric uncertainties in the system model. This paper presents a decentralized robust prescribed performance control approach. It is inherently robust to the system nonlinearities, the actuator failures, and the disturbances. It exhibits lower costs in computation, higher efficiency in design, and is more user-friendly in implementation. It achieves trajectory tracking with preassigned rate and accuracy, despite the actuator failures. Extension of the approach to multi-agent systems with actuator failures is an interesting topic for future investigations.