【作者】 高畅；

【导师】 覃亚伟；

【作者基本信息】 华中科技大学 ， 建筑与土木工程， 2021， 硕士

【摘要】 近年来我国道路建设取得巨大成就,交通压力却不断提升。我国在向着“交通强国”目标迈进的同时,发布了多项土木行业信息化、自动化升级的文件。道路施工方案作为每个道路项目技术与组织上必不可少的指导文件,目前其传统的经验编制方法仍具有较大改造空间。本文使用案例推理技术进行道路施工方案的快速智能生成,并利用云推理、贝叶斯网络等理论在案例检索与方案应用中采取了多项优化措施。主要研究内容如下:(1)分析道路施工技术、道路施工方案内容等知识特点,将道路施工方案知识分解为案例特征、解决方案两大部分。然后根据案例检索技术需要,提出其各个细分指标的数据存储形式并说明取值范围,形成了完整的道路施工方案案例库结构形式。(2)案例检索过程中利用KNN(K-Nearest Neibghor)算法进行案例间的相似度计算,针对各类型数据间距计算提出了合理计算方式。然后提出了基于二维云推理的权重优化方法以削弱案例检索时模糊输入值的影响,实现了整个道路施工方案智能生成。在方案输出后的应用过程中,本文构建道路质量风险贝叶斯网络,基于现场信息对道路的质量风险相关问题进行分析并确定生成方案的改进目标。(3)以南方某城市某已竣工公路项目为例,构建案例库并使用CBR技术快速检索最似案例,修正后获得了更优的施工方案。然后,使用二维云推理技术对KNN算法中属性权重进行优化,获得了更贴合案例实际特征的案例相似度计算结果。最后,通过风险贝叶斯网络,以原始方案施工中的相关质量检测数据计算得到该项目施工前重点质量关注指标、过程中的质量风险大小,并确定了原始方案的改进方向。其结果显示该项目原始方案待改进章节与上述CBR生成方案优化的章节基本一致。本文结合实际工程案例,证实了基于CBR的道路施工方案智能生成方法的可行性与适用性,为道路施工方案编制自动化及提升方案应用效果提供了新思路。更多还原

【Abstract】 Our country has made great achievements in road construction in recent years,but the traffic pressure has continued to increase.While striding forward to the goal of being a "transportation power",our country has published a number of documents for the informatization and automation of the civil engineering industry.The road construction plan is an indispensable guidance document for each road project in terms of technology and organization.At present,its traditional experience preparation method has a lot of room for improvement.This thesis used case-based reasoning technology to quickly and intelligently generate road construction plans,and used cloud reasoning,Bayesian networks and other theories to take a number of optimization measures in case retrieval and plan application.The main research contents were as follows:(1)Analyzed the knowledge characteristics of road construction technology and the content of road construction plans,and decomposed the knowledge of road construction plans into two parts: case characteristics and solutions.Then,according to the technical needs of CBR,the data storage form of each sub-index was proposed and the value range was explained,forming a complete road construction plan case library structure.(2)In the case retrieval process,the KNN(K-Nearest Neibghor)algorithm was used to calculate the similarity between cases,and a reasonable calculation method was proposed for the calculation of various types of data spacing.Then a weight optimization method based on two-dimensional cloud reasoning was proposed to weaken the influence of fuzzy input values in case retrieval,and realized the intelligent generation of the entire road construction plan.In the application process after the program output,this thesis constructed a road quality risk Bayesian network,analyzed the road quality risk-related issues based on on-site information,and determined the improvement target of the generated program.(3)By taking a completed highway project in a southern city as an example,constructed a case database and used CBR technology to quickly retrieve the most similar cases,amended to obtain a better construction plan.Then,two-dimensional cloud inference technology was used to optimize the attribute weights in the KNN algorithm,and the case similarity calculation results that were more suitable for the actual characteristics of the case were obtained.Finally,through the risk Bayesian network,the relevant quality inspection data during the construction of the original plan was used to calculate the key quality concern indicators before the construction of the project,the quality risk in the process,and the improvement direction of the original plan was determined.The results showed that the chapters to be improved in the original plan of the project were basically the same as the chapters on the optimization of the CBR generation plan.Based on actual engineering cases,this thesis proves the feasibility and applicability of the intelligent generation method of road construction plan based on CBR,and provides new ideas for the automation of road construction plan preparation and improvement of plan application effects.更多还原