Self-Healing Control for Large Launch Vehicle Based on Extended State Observer and Adaptive Dynamic Programming

This paper investigates the self-healing control problem for the large launch vehicle (LLV) with system uncertainties, external disturbances, and actuator faults. First, the attitude dynamics of the LLV are presented, and the control-oriented model subject to undesired malfunctions is structured. Second, a novel extended state observer (ESO) is designed to estimate the disturbances and fault informations, and the nonlinear gain functions with two different linear ranges are introduced to improve the estimation accuracy and reduce the effect of peaking value problem. Then, an integral terminal sliding mode fault tolerant control scheme is proposed for the attitude faulty system, which would stabilize the closed-loop system even existing the disturbances and actuator failures. Besides, utilizing the adaptive dynamic programming (ADP) technique, a supplementary control with actor-critic structure is employed to further improve the system tracking performance and provide the additional compensation control input according to the bias between the desired value and actual one. Finally, the effectiveness of the proposed method is verified by the simulation results.