摘要
对自组织神经网络改进算法的收敛性进行了证明。当该算法中的参数和网络权值初始值选取适当小的正实数时，网络收敛。文中并分析了该算法的收敛性与网络参数、初始值之间的关系。
The convergence of the improved algorithm based on the self-organization neuralnetwork for dynamic nonlinear system identification is proved in this paper. The algorithm isstable if the network has suitable initial weight values and low positive-real parameters. Thenthe relation between convergence of the network and the initial values of the network is analyzed.
出处
《电机与控制学报》
EI
CSCD
2000年第4期236-238,共3页
Electric Machines and Control
关键词
神经网络
自组织特征映射
收敛性
neural network
self-organization feature map
convergence
