摘要
前向神经网络已在设备的故障诊断中得到普遍应用。神经网络的性能和模式识别能力与网络的拓扑结构和学习算法有着密切的联系,文章研究并实现了在故障诊断应用中能进行自构形的前向神经网络模型,改善了全局收敛性和节点总体饱和度,并较好地应用于电气设备故障诊断实践中。
Feedforward Neural Network(FNN)has been extensively used to diagnose equipment faults.FNNs' performance and pattern-discriminating capacity have relation with their topological structure and training algorithm.This paper stud-ies FNNs' auto-structural model in diagnosing faults,for improving the overall convergence and nodes' collectivity satu-ration.This model has been applied in electric faults detection.
出处
《计算机工程与应用》
CSCD
北大核心
2004年第20期227-229,共3页
Computer Engineering and Applications
关键词
神经网络
自动构形
模式识别
故障诊断
Artificial Neural Network,auto-structure,pattern recognition,fault diagnosis
