Multi-state Reliability Analysis Of A Large Deployable Satellite Antenna Based On Bayesian Networks

Since the deployment system is one of the most important parts in large satellite antenna, its analysis and prediction of the reliability is essential for the satellite. In this paper, we use Bayesian Networks(BN) to analyze the multi-state reliability of a large satellite antenna deployment system. Firstly, introduced the concept and basic principles in Bayesian Networks, describe the bucket elimination algorithm theory and implementation of reasoning process detailedly. Secondly, established the BN model of a multi-state large satellite antenna deployment system, which means each node in the BN model has three work states-"normal","critical","failure". In the case of given the prior probability of the root nodes, use the independent relations and bucket elimination to calculate the conditional probability of the rest nodes in BN, and analysis the importance of the root nodes. Then, use the random-Fuzzy method to analysis the key failure event of antenna deployment system, and calculated the fuzzy possibility of the performance function. Finally, give a brief description of the Satellite Antenna System Reliability Prediction Software, which is based on the satellite antenna deployment system's BN model, bucket elimination algorithm and random-Fuzzy method.