摘要
研究了自组织特征映射 (SOM)神经网络的结构及学习算法 ,提出了利用SOM神经网络对输入样本的“聚类”作用 ,实现对故障模式的分类 ,并据此对故障进行诊断的新方法。通过对某卫星电源系统故障的诊断实例 ,验证了该方法是有效可行的 ,能对卫星电源系统的故障进行准确的实时诊断。
The construction and algorithm of a self organizing feature map neural network is researched.A new fault diagnosis method based on clustering principle of the SOM network is presented and its feasibility is proved by diagnosing the faults of satellite power system successfully.All kinds fault of satellite power system can be real time diagnosed.
出处
《强度与环境》
2002年第2期38-41,共4页
Structure & Environment Engineering
基金
航天科技重点课题
国防"九五"预研项目基金资助 (4 9.1.2 )
