Indirect neural-based finite-time integral sliding mode control for trajectory tracking guidance of Mars entry vehicle

Abstract

This article presents an indirect neural-based finite-time integral sliding mode control algorithm for the reference trajectory tracking guidance of Mars entry vehicle under uncertainties. The proposed controller is developed as a combination of finite-time integral sliding mode controller and indirect neural identification. The finite-time integral sliding mode controller is designed by constructing a new type of finite-time integral sliding mode surface to prevent the singularity problem. Moreover, the neural network (NN) is combined with the finite-time integral sliding mode controller to identify the lumped uncertainty and attenuate the chattering phenomenon. Particularly, the concept of indirect neural identification is adopted and only a single adaptive parameter is required to be learned online. In this way, the proposed controller is not only strongly robust against aerodynamic and density uncertainties, but also computationally simple for onboard implementations. Stability argument indicates that the proposed controller can ensure the radial distance tracking error and its time differentiation regulate to the small residual sets around zero in finite time. Lastly, the effectiveness and excellent guidance performance of the proposed control algorithm are demonstrated through simulation studies on a Mars Science Laboratory-type (MSL-type) entry vehicle.