基于信息扩散和BP网络的黄土边坡稳定性分析  

The evaluation of loess slope stability based on combination of information diffusion theory and BP neural metwork

作　　者：何书[1,2] 王家鼎[1] 王欢[3] 韩晓萌[1] 

机构地区：[1]西北大学地质学系／大陆动力学国家重点实验室,陕西西安710069 [2]江西理工大学资源与环境工程学院,赣州341000 [3]西安市勘察测绘院,陕西西安710054

出　　处：《西北大学学报（自然科学版）》2008年第6期983-988,共6页Journal of Northwest University（Natural Science Edition）

基　　金：国家自然科学基金资助项目(40572157);高等学校博士点专项科研基金资助项目(20050697013)

摘　　要：目的探讨基于信息扩散原理的BP神经网络的黄土边坡稳定性评价模型。方法收集黄土地区24组黄土边坡实例,采用模糊信息优化处理中的信息扩散原理,建立各评价因子与安全系数之间的模糊关系,并在此基础上建立与BP神经网络相结合的评价模型。结果建立的评价模型对4组预测样本的预测结果,效果良好,较好地解决了样本过少或含有矛盾样本的问题。结论该模型在黄土边坡稳定性评价中比普通神经网络具有更高的实用性和有效性。Aim To establish an evaluation model of the prediction of the slope stability based on combination of information diffusion theory and BP neural network. Methods The 24 sets of slope data in loess area were collected. The fuzzy relationships between every evaluating elements and safety coefficient were established by the infor- mation diffusion principle. An evaluation model was established based on combination of information diffusion theory and BP neural network and this model can well resolve the phenomenon of less practical samples, the existing contradictory samples. Results The model was verified by 4 series data of collapsing loess and the result is very good. Conclusion The result shows that the model is more useful and valid than the common neural network.

关 键 词：信息扩散 黄土 边坡稳定性 BP神经网络 

分 类 号：P642.131[天文地球—工程地质学]