Research On Key Autonomous Technology Of Spacecraft System Based On Multi-Agent

In the new century ,space exploration has become a hot topic of the aerospace field. In space exploration, an important aspect of the research is autonomous technology. Its goal is to control and accurately apperceive their own state and the external environment not rely on outside information and control or minimize the dependence on the outside world, in the light of these information and user tasks to make all the appropriate decisions, and able to independently control the spacecraft to complete various tasks. In the study of the problem, there were two key technologies .The architecture of autonomous system and autonomous planning and scheduling algorithm. This paper focus on these two key technologies, and describe it from four-parts: research on the architecture of autonomous system, research on planning and scheduling algorithm , the demonstration of autonomous system, the expansion study about Intelligent algorithm.The first part is the architecture of autonomous system, multi-Agent system is widely used in various fields, In this paper, A specific multi-Agent architecture based on the former research is raised, and implemented under VxWorks operation system, including the formal description of satellite subsystems, each function and state of satellite, the information feedback and control mechanism and the mechanism of communication.The second part is the autonomous planning and scheduling algorithm, which is based on characteristics of spacecraft and mission target. In this part, some basic conceptual model is described, for example, the activities, resources and constraints. and so on, and combine several related sequence optimization theory, raise a suitable model for the space task planning and scheduling, then focus on the analysis of mathematical model, which is divided into two parts: offline and online. The off-line scheduling is the focus, and the comparative analysis of a wide range of intelligent algorithm is implemented, at last, The conclusion is that genetic algorithm, simulated annealing is the ideal algorithm for offline scheduling. For online scheduling, heuristics algorithm is applied.The third component is a simple demonstration of the autonomous system. The demonstration simulates the entire external systems and internal systems. External system contains ground stations, spacecraft, and the communication between these two parts. Internal system contains startup of the target machine, VxWorks OS , and the autonomous software. In autonomous software ,some function is achieved, including self-control mechanism, feedback mechanisms, and the simplify planning and scheduling mechanism, and finish the simulation of task_excution and failure_dealling.The fourth part is the expansion research of intelligent algorithm, which is based on the second-part: scheduling algorithm research. Primarily about the many intelligent algorithm (genetic algorithm, simulated annealing, ants algorithms, pso algorithm, and some hybrid algorithm). In light of the experimental results and the features of the subject, The research is done on the intelligent algorithm, which includes choice of algorithm parameters, features of the algorithm. Finally on this basis, General conclusions is reached for the mathematic model with intelligent algorithm.This paper mainly describe the architecture of autonomous system and the planning and scheduling algorithm technology of autonomy. Multi-Agent Architecture demonstration experiment has proved that the architecture is feasible in space application under current technical conditions. In addition, through experimental research on planning and scheduling algorithm, particularly off-line scheduling algorithm research, has proved that intelligent algorithm is efficient for the space off-line scheduling. Hope that the results of this paper make certain contributions to the development of the autonomous techniques.