Control Of Deployable Antenna Based On Genetic Algorithms And Artificial Neural Networks

This paper belongs to a thesis called the study on dynamical analysis and control of large space deployable antenna which sponsored by the National Natu-ral Science Fund.In the area including communications, Earth observation, space science and technology, large space antennas were made use of more widespread. Limited by the room in the spacecraft, the antenna must be folded back in on itself when the rocket launching before the antenna spreads to its work shape depending on it's own power in position. Therefore, deployable antenna has become one of the main developing trends of modern space technique. Many states contest to de-veloping this technology.Because of such characters as large size, light weight, and great flexible,the structural flexibility of large space deployable antennas should no longer be ignored when moving. And there is a strong coupling be-tween the flexible deformation and rigid displacement of the body. Therefore, there are many difficulties in the dynamical analysis,structure design and control of such a complicated nonlinear system. As a result, it has attracted widespread attention of many experts and scholars inside and outside the country.In this paper, the following has mainly been done:1. Based on the theory of Flexible multi-body dynamics, the model of deploy-able antenna is established, and the dynamic response and analysis calcula-tion to large space deployable antenna is done.2. Discussed in detail the origin,the development and application of methods of the theory of genetic algorithms, outlined the principles and application of artificial neural networks.3. The idea of ANN's disassembly and composition has been proposed, and by combining the idea with genetic algorithm theory, the three different forms of system identification were done, their advantages and disadvantages were compared. The results show that the model of genetic algorithm combining disassembly ANN is greater accuracy than the model of simple identification ANN or the model of genetic algorithm combining complex ANN. Using the disassembly model, the control system--- ANN predictive scrolling optimiz-ing control system applied to the speed control of deployable Antenna is de-signed. Practice has proven that the idea of ANN's disassembly and compo-sition has a great prospect and some reference value, in optimization or con-trol of the huge multi - variable nonlinear systems.4. The computer simulation results show that the control system is not only to meet the requirements of control, but also is better than the control system from a simple neural network model on accuracy, agility, robustness.