Spacecraft attitude estimation with the aid of Locally Linear Neurofuzzy models and multi sensor data fusion approaches

In this paper the locally linear neurofuzzy (LLNF) models with data fusion approach are used to solve the spacecraft attitude estimation problem based on magnetometer sensors and sun sensors observations. LLNF with locally linear model tree (LoLiMoT) algorithm as an incremental learning algorithm have been used several times as a well-known method for nonlinear system identification and estimation. The efficiency of the LLNF estimator is verified through numerical simulation of a fully actuated rigid body with three sun sensors and three-axis-magnetometers (TAM). For comparison, Kalman filter (KF) as a well-known method in spacecraft attitude estimation and MLP and RBF neural networks are used to evaluate the performance of LLNF. The results presented in this paper clearly demonstrate that the LLNF is superior to other methods in coping with the nonlinear model.