基于实例推理的堤防防渗加固方法选择  

Optimization of Reinforcement Strategies for Embankment Based on Case- Based Reasoning

作　　者：张林海[1,2] 苏怀智[1,2] 李皓[2,3] 

机构地区：[1]河海大学水文水资源与水利工程科学国家重点实验室,江苏南京210098 [2]河海大学水利水电学院,江苏南京210098 [3]水资源高效利用与工程安全国家工程研究中心,江苏南京210098

出　　处：《人民黄河》2015年第1期46-49,53,共5页Yellow River

基　　金：江苏省杰出青年基金资助项目(BK2012036);高等学校博士学科点专项科研基金资助项目(20130094110010);国家自然科学基金资助项目(51179066;41323001);水利部公益性行业科研专项(201301061;201201038)

摘　　要：针对堤防工程防渗加固方法的优选问题,充分借助已有工程的加固案例,综合应用基于实例的推理(Case Based Reasoning,CBR)方法和自组织特征映射神经网络(SOFM),在对影响加固方法选择的主要因素分析基础上,研究建立加固工程和实例之间相似性的计算公式;利用自组织特征映射神经网络高度的自组织性和自适应性,对实例库中的实例进行动态聚类,提出以SOFM为检索机制的实例检索模型。将所述模型和方法应用于某实际工程,研究该堤防工程防渗加固方案生成的过程,分析模型和方法的有效性。算例分析表明:利用所建模型能够缩小实例检索的范围,提高检索的效率;有效利用了以往加固实例中积累的经验和知识,减少了对专家的依赖,可提高加固方法选择的效率和加固决策的智能化水平。Based on the Case Based Reasoning （CBR）integrated with Self Organizing Feature Map （SOFM）,the selection of seepage con-trol measures for dyke reinforcement was discussed in this paper. To make full use of existing reinforcement cases,after analyzing the main factors affecting the selection of seepage control measures,a formula computing the similarity between the reinforcement case and the existing case in the case library was set up. Due to the high quality of SOFM on self-organization and self-adaptation,which was adopted in the dynamic clustering for cases in the case library,a case retrieval model based on SOFM was presented. One real dyke reinforcement case was given to illustrate the working process of the model,after that the effectiveness of the model was analyzed. The case shows that the use of the model can narrow the retrieval range and improve retrieval efficiency. Meanwhile,because of the effective use of the experience and knowl-edge accumulated in the past reinforcement cases,the model can not only reduce the reliance on experts,but also improve the efficiency on reinforcement method selection and the intelligence level of decision-making.

关 键 词：堤防工程 防渗加固 方案选择 实例推理 自组织特征映射神经网络 

分 类 号：TV871[水利工程—水利水电工程]