摘要
在分析BP神经网络建模步骤的基础上,针对BP神经网络某些不足,提出了几点改进措施。首先对原始数据进行了非线性规格化;其次,提出了记忆式初始权值和阀值;最后以确定性系数最大为依据进行参数优选,并将改进后的BP神经网络应用于需水量预测。结果表明,改进后的BP神经网络不仅提高了BP神经网络预测的精度,而且加快了BP网络运行时的收敛速度。
Based on analyzing the procedure to establish the model for BP Neural Network and aimed at some deficiencies that exit in BP Neural Network, several improvements to BP Neural Network are come out. At first, original data are non-linear regularized, then remembrance preliminary weight and valve are put forward, and at last parameters are chosen on the basis of the maximum determining coefficient. And the improved BP Neural Network is applied to water demand prediction. The case shows that the improved BP Neural Network not only can improve prediction precision but also can expedite convergence pace.
出处
《河南科学》
2003年第2期202-206,共5页
Henan Science
基金
河南省自然科学基金项目(004040600)
