摘要
介绍广义回归神经网络(GRNN)的原理和影响因素,论述光滑因子的影响和选择。采用LOO交叉验证方法遍历所有样本,搜索出合适的光滑因子,结果表明合适的光滑因子能够较大幅度地提高网络泛化能力。应用收集到的82个圆弧滑面边坡稳定状态的实例资料,将GRNN模型应用于边坡稳定性评价,计算结果表明,在边坡稳定状态分析及预测方面,GRNN模型比BPNN模型更加精准简捷。
The principles and influence factors of the algorithm of generalized regression neutral network（GRNN） were introduced.The influence and choice of the spread were discussed.The suitable value of the spread was searched out from all the samples according to the LOO across validity method.It was proved that the suitable value of the spread could greatly improve the generalization ability of the network.The method of GRNN was applied to the evaluation of slope stability of 82 examples.The results show that the GRNN model is simpler and more accurate than the BPNN models in prediction and analysis of slope stability.
出处
《水利水电科技进展》
CSCD
北大核心
2011年第3期80-83,共4页
Advances in Science and Technology of Water Resources
基金
国家自然科学基金(50979030
50911130366)
关键词
广义回归神经网络
光滑因子
边坡稳定预测
generalized regression neural network
spread
prediction of slope stability
