论文部分内容阅读
及时发现在轨卫星的异常状态是卫星监控的一项重要工作,它不仅关系到卫星的正常使用,也影响卫星的寿命.为了能够及时发现卫星的故障,并且从中提取异常模式,提出一种基于Prefix Span算法的卫星异常模式挖掘方法.首先通过灰关联分析剔除冗余参数,然后利用各参数阈值提取异常数据集并且通过信息熵对各参数进行离散化,最后使用Prefix Span算法对异常数据集进行模式挖掘.该方法在数据量大、信息复杂的前提下,实现了对卫星异常模式的挖掘.通过对某卫星数据的实验分析,验证了所提方法的可行性和有效性.
It is an important task of satellite monitoring to find out the abnormal state of orbiting satellites in time.It not only affects the normal use of satellites but also affects the life of satellites.In order to detect satellite faults in time and extract abnormal patterns therefrom, Prefix Span algorithm for satellite anomaly pattern mining method is proposed.First, the redundant parameters are eliminated by the gray relational analysis, and then the abnormal data sets are extracted by using the thresholds of each parameter and the parameters are discretized by the information entropy. Finally, Mode mining.This method realizes the excavation of satellite anomaly patterns on the premise of large amount of data and complex information.Through the experimental analysis of a satellite data, the feasibility and effectiveness of the proposed method are verified.