科研之友
微信
新浪微博
Facebook
分享链接
摘要
Stellar refraction-based celestial navigation is an attractive method with high accuracy and low cost. Most existing research works only focus on orbit determination problems for orbital vehicles in outer space while its working domain is limited. In order to expand its applicable area and improve information utilization, a new fault-tolerant stellar refraction-based inertial/celestial integrated navigation system is designed in this work, which is supposed to provide accurate position, velocity, and attitude information for aerospace vehicles that either make a maneuvering flight in near space or move in a predetermined Earth orbit in outer space. First, a new nonlinear navigation system dynamic model is established by error-prorogation equations of the strapdown inertial navigation system (SINS) in the Earth-centered inertial (ECI) frame, in which additive quaternion with less model error is used for attitude computation. Secondly, celestial navigation subsystem (CNS) is used to indirectly sense the horizon by utilizing the starlight atmospheric refraction model and determine the vehicle attitude by means of a multistar vector observation method. Thirdly, a fault-tolerant federated unscented Kalman filtering (FFUKF) algorithm is employed to perform information fusion, aiming at improving the navigation accuracy and reliability. Numerical simulations are conducted for an aerospace vehicle either in a near space maneuvering flight or in an orbital flight along a low Earth orbit. The results show that the proposed strategy achieves higher accuracy than traditional methods, and all the navigation parameters can be estimated. Moreover, different measurement faults are introduced in the simulations, and the FFUKF algorithm manages to detect and isolate them and keep the navigation system working in normal operation.
