基于贝叶斯网络的跳频序列多步预测  

Multi-step prediction of frequency-hopping sequences based on Bayesian networks

作　　者：张恒伟[1] 杨有龙[1] 朱原媛[1] 

机构地区：[1]西安电子科技大学理学院,西安710071

出　　处：《计算机应用研究》2012年第1期237-240,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61075055)

摘　　要：根据跳频频率序列具有混沌特性,在相空间重构理论基础上提出一种用于跳频频率序列预测的贝叶斯网络模型。该模型将重构后的整个相空间作为先验数据信息,进而通过学习贝叶斯网络并利用贝叶斯网络推理算法达到对跳频频率多步预测的目的。仿真结果表明该方法具有良好的多步预测能力,并能有效地克服过拟合现象。According to the chaotic characteristics of frequency hopping（FH） sequences and based on the phase-space reconstruction,This paper proposed a Bayesian network model applied to FH sequences prediction.The model regarded entire phase-space as a priori data information to learn Bayesian networks.Then it achieved the multi-step prediction of frequency hopping using the Bayesian network inference algorithm.Experiment results show that the proposed method can provide high accurate precision with steady multi-step prediction ability and avoid the over-fitting effectively.

关 键 词：跳频序列 贝叶斯网络 相空间重构 多步预测 

分 类 号：TN975[电子电信—信号与信息处理]