摘要
首先提出 RBF隐层节点冗余结构 ,然后分别分析在单故障和普遍故障下的网络特性 ,并证明了这种结构可以有效容错 ,最后提出了实用的隐层冗余方法 ,并经仿真实验证明是可行的。
A hidden-layer neuron redundancy architecture of RBF network is presented. Performance of the network are analyzed under both single-fault and universal faults. The performance of the network under faults can be improved significantly with this redundancy architecture. Finally, a practical method of redundancy of hidden-layer neurons to gain fault tolerance is presented. Simulation results show the validity of the method.
出处
《控制与决策》
EI
CSCD
北大核心
2001年第5期591-594,598,共5页
Control and Decision
关键词
容错
冗余
多层感知器
径向基神经网络
学习算法
Algorithms
Architecture
Computer simulation
Fault tolerant computer systems
Fault tree analysis
Redundancy
Theorem proving
