论文部分内容阅读
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性,又能大大加快其收敛速度。
This paper discusses the effect of BP algorithm on training rate when selecting system number and establishes a criterion to select the number of learning samples at each moment. Under this standard, not only the relativity of training network but also the acceleration of convergence speed.