Intelligent Sensor Fusion in High Precision Satellite Attitude Estimation Utilizing an Adaptive-Network-Based Fuzzy Inference System

In this study, Adaptive Network-Based Fuzzy Inference System (ANFIS) is presented with sensor data fusion approach to estimate satellite attitude. The active sensors are sun and earth sensors. Satellite attitude dynamic, including attitude quaternion and angular velocities are estimated simultaneously utilizing the measured values by the sensors. The Extended Kalman Filter (EKF) is employed to verify and evaluate the efficiency of the presented method. Additionally, the neural networks with Radial Basis Function (RBF) and Multi-Layer Perceptron (MLP) are also designed to prove the superiority of the proposed ANFIS network among the smart methods of sensor data fusion for satellite attitude estimation. Root Mean Square Error (RMSE) as a numerical criterion and graphical analysis of residues are utilized to evaluate the simulation results. The simulations confirm that the obtained estimations from ANFIS network have more accuracy in modeling of nonlinear complex systems compared to EKF, MLP and RBF networks. In general, using intelligent data fusion, especially ANFIS, reduces attitude estimation error and time in comparison to the classical EKF method.

Keywords:

attitude estimation , Data Fusion , ANFIS , Extended Kalman Filter , Neural Network