Reliability-guaranteed anti-disturbance control via closed-loop uncertainty quantification

Abstract

In order to quantitatively manage the failure probability/reliability of control systems, we propose a reliability-guaranteed anti-disturbance control (RADC) method via the closed-loop uncertainty quantification (CL-UQ). Specifically, multiple probabilistic intervals are proposed to model the parameter and disturbance uncertainties. Through the CL-UQ analysis, the uncertainty characteristic of the system states can also be depicted by multiple probabilistic intervals. Combining with the designed prescribed performance threshold, a time-varying reliability model for the control process is established. On this basis, the control optimization problem with a reliability constraint for the RADC is formulated to obtain the optimal feedback gain and observer gain. A simulation example of spacecraft attitude stabilization shows that the CL-UQ results meet with the high accuracy and the RADC effectively reduces the conservatism of the traditional anti-disturbance control.