摘要
讨论BP算法用于系统辨识时样本数的选取对训练速率的影响。确立一种选取每一时刻学习样本数的标准 ,在此标准下 ,既能保证训练网络的相对稳定性 。
The influence of selecting the proper number of samples on learning pace when BP algorithm is used for system identification is discussed in this paper.A standard of selecting the number of samples at each time is proposed.Under the standard,BP neural network is not only stable but also of very fast convergent speed.
出处
《武汉科技大学学报》
CAS
2001年第2期183-185,共3页
Journal of Wuhan University of Science and Technology
