基于概率神经网络的相似度排序方法  

A Similarity Ordering Method Based on Probability Neural Network

作　　者：崔伟成[1] 刘持胜[2] 孟凡磊[1] 刘林密[1] 

机构地区：[1]海军航空工程学院飞行器工程系,山东烟台264001 [2]海军航空工程学院外训系,山东烟台264001

出　　处：《海军航空工程学院学报》2014年第6期593-596,共4页Journal of Naval Aeronautical and Astronautical University

基　　金：学院青年科研基金项目资助(HYQN2013-02)

摘　　要：为了研究相似度排序问题,对概率神经网络的结构进行了分析,对其输出进行了改进,使之输出后验概率,进而进行相似度排序。结合电火工品发火性能的预测问题,对方法进行了验证,结果表明:基于概率神经网络的相似度排序方法与实际结果相符,方法是可行的。A similarity ordering method based on probability neural network was studied. The output of the probability neu- ral network was chosen as the posterior probability based on the analysis of the structure. The posterior probability could be used as the warrant of the similarity. In order to validate the method, the predict of the firing sensitivity of the electric explosive device was put forward. The result showed that the predicted result was consistent with the measured resuh. The method was authentic.

关 键 词：概率神经网络 相似度 后验概率 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]