Nonlinear Optimal Guidance Control for Lunar Trajectory Tracking Descent
- Paper number
IAC-05-C1.8.09
- Author
Dr. Dayi Wang, Beijing Institute of Control Engineering, China
- Coauthor
Prof. li tieshou, Beijing Institute of Control Engineering, China
- Year
2005
- Abstract
The lunar soft landing in this paper begins from a circular lunar parking orbit. Once the landing area has been selected and it is time to de-orbit for landing, a burn is performed to establish an elliptical orbit. At perilune, the landing jets are ignited, and a propulsive landing is performed to make the horizontal velocity relative to the lunar surface nearly zero. Then a vertical landing is taken by on/off control of thrusters at the final phase. Most of fuel is consumed in powered descending phase, therefore the guidance law is important for the design of a minimal fuel landing. A nonlinear optimal on/off guidance law for trajectory tracking descent is proposed in the paper. Optimal control theory is used to generate a guidance law for lunar soft landing, which can be regarded as an open loop one. Generally, a two-point boundary value problem (TPBVP) is to be solved in order to obtain the numerical solution for optimal trajectory. Shooting methods based on an initial value guess technique are proposed to solve the TPBVP. By regarding the optimal trajectory as a nominal one, a closed-loop nonlinear guidance law is synthesized on the basis of neural networks’ attractive properties for nonlinear problems. In this paper, we investigate the use of neural networks as a guidance controller in the whole phase of landing from perilune to the surface on the moon. We utilize an improved BP algorithm to train the neural network to learn to function as a closed-loop soft landing control system and to force the dynamics of the lander to match that of a specified optimum trajectory. Simulation results show that a sub-optimal soft landing trajectory can be obtained by the neuro-guidance law even with a certain initial navigation errors and measurement errors, and also shed light on the ability of the neural network design technique to serve as a potential tool for design and analysis of spacecraft guidance law. Clearly, the synthesis process presented in this paper can be used to exploit specific landing design concepts. More complex target maneuvers must be investigated, and due to limited speed of the GNC computers and time constraints for landing, the neural network training has to be carried out off-line.
- Abstract document
- Manuscript document
IAC-05-C1.8.09.pdf (🔒 authorized access only).
To get the manuscript, please contact IAF Secretariat.