BP和RBF神经网络在边坡稳定性评价中的应用  

The Application of BP and RBF Neural Networks in Slope Stability Evaluation

作　　者：张玉灯[1] 郑俊[1] 

机构地区：[1]三峡大学三峡库区地质灾害教育部重点实验室,443002

出　　处：《灾害与防治工程》2008年第1期47-52,共6页Disaster and Control Engineering

基　　金：国家自然科学基金(50379023)

摘　　要：基于BP网络和RBF网络理论,选取影响岩质边坡稳定性的一些主要因素,建立了边坡稳定性分析的神经网络模型,并用Matlab7.0神经网络工具箱对一些边坡样本进行训练仿真。对比了两种网络的逼近精度和预测结果,结果表明:两种网络均可以用于边坡的稳定性评价,RBF网络的性能要优于BP网络,网络最优参数的选择要通过反复实验获得。In this paper, the neural analyzing model has been constituted based on BP and RBF Neural Network theories, and the factors that influence the slope stability have been taken into account. The network is then trained and simulation based on slope samples data set by use of the NN toolbox in Matlab 7.0. Comparison between two Networks in the approximate accuracy and prediction results is conducted. The results show that two Neural Networks are both suitable for assessing the stability of rock slope, the RBF model is better than the BP model and the Optimal Selection on the Parameters of Neural Network can be achieved by experiments.

关 键 词：BP神经网络 RBF神经网络 边坡稳定性 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] U416.13[自动化与计算机技术—控制科学与工程]