Mass properties identification and automatic mass balancing system for satellite attitude dynamics simulator

Author(s):  
Ghasem Sharifi ◽  
Mehran Mirshams ◽  
Hamed Shahmohamadi Ousaloo

A Satellite Attitude Dynamics Simulator is a low-cost, ground-based system made to simulate the conditions of a weightless satellite in space. The identification of the mass characteristics is crucial for Satellite Attitude Dynamics Simulator application and so the center of mass place is necessary for balancing the platform and moment of inertia which is a significant factor in designing controllers and selecting actuators. The purposes of this paper are the mass properties identification and design, experimentation, and validation of an automatic mass balancing system, which is assembled on the Satellite Attitude Dynamics Simulator at the Space Research Laboratory. This paper presents a process of mass properties estimation for the Satellite Attitude Dynamics Simulator using classical Levenberg–Marquardt as an optimization method. By employing this technique lack of repeatability and difficulties in implementation will be eliminated. In order to verify this technique, a MATLAB® SIMULINK® model of the Satellite Attitude Dynamics Simulator is established. The gap between the center of mass and center of rotation is decreased by means of the automatic mass balancing system in order to remove gravity disturbance. The results of this identification process are compared to the recursive least square algorithm, which is commonly employed in identification of mass properties. The analytical and experimental results prove that the proposed characteristic estimation process using classical Levenberg–Marquardt algorithm is more effective and appropriate. Proper excitation of the platform will guarantee the accuracy of estimation and compensation of the center of gravity offset utilizing the balancing system.

1976 ◽  
Vol 13 (1) ◽  
pp. 39-64 ◽  
Author(s):  
L. G. Kraige ◽  
J. L. Junkins

10.29007/q7pr ◽  
2019 ◽  
Author(s):  
Ana Farhat ◽  
Kyle Hagen ◽  
Ka C Cheok ◽  
Balaji Boominathan

Electronic Brake System (EBS) is considered as one of the most complicated systems whose performance depends on the subsystems parameters. Usually these parameters are difficult to predict. Based on the task to improve the EBS performance, this article presents a mathematical modeling approach based on neuro-fuzzy network method to model a subsystem of EBS. For the model parameters identification, a neuro-fuzzy network has been implemented based on Least Square Error (LSE) and Levenberg- Marquardt Algorithm (LMA) as the optimization algorithms. Finally, the performance of identified model has been evaluated.


Sign in / Sign up

Export Citation Format

Share Document