【作者】 冯仲仁；

【导师】 朱瑞赓；

【作者基本信息】 武汉理工大学 ， 结构工程， 2003， 博士

【摘要】 上世纪90年代以来，我国高速公路发展迅速，每年新建高速公路2000公里左右，到2002年底已建成高速公路总里程22500多公里，超过加拿大稳居世界第二位。我国许多省市的公路地基中大量存在着淤泥、淤泥质粘土、淤泥混砂层。这些地基大多压缩性高，沉降量大，排水固结稳定性差。在其上修建高速公路，若不经过处理或处理不当，会引起道路质量的降低甚至破坏。随着高速公路的修建，软土问题越来越突出，已成为影响工程质量，工程周期和工程造价的关键因素之一。因此，加强高速公路软基处理的研究，已经成为影响我国交通事业的发展的一个非常重要课题。 在高速公路软基处理的技术研究中，存在着大量的决策问题。在设计阶段，首先需要对是否进行处理做出决策，对需要处理的地段，则需要对深层处理还是浅层处理做出决策。对于浅层处理或深层处理的具体方法也需要做出决策。对于某一种处理方法，需要对其技术参数做出决策。在施工阶段，需要根据动态监测的结果，对施工期的加载速度做出决策。根据工后沉降的预测结果，对路面的加铺时间做出决策。 在软基处理技术决策中，影响因素众多，计算工作量大，设计理论不完善，经验往往起重要的作用。因此，如何使软基处理技术决策更具科学性变得非常重要。本文以近十年高速公路软基处理的工程实践为基础，深入地研究了高速公路软基处理的诸多决策问题，明确了当前高速公路软基处理决策中存在的有待解决的课题。以系统工程、人工智能、知识工程、综合评价、神经网络、模糊数学、遗传演化算法、决策分析、计算机技术等为理论基础，结合岩土工程、土质学和土力学等学科的最新发展，对建立高速公路软基处理智能决策支持技术的各方面问题进行了研究，为高速公路软基处理决策的智能化和科学化进行了有益的探索，填补了这方面的空白。 按着研究内容和论文的章节顺序，简要介绍如下： 1.简要地介绍了本文研究工作的目的和意义，较全面地介绍了目前软基处理的研究现状，存在的问题和今后的研究展望。在系统分析了高速公路软基处理中的诸多决策问题的基础上，提出了高速公路软基处理智能决策支持系统的研究题目和主要研究内容。 2.将人工神经网络(ANN)用于高速公路软基处理的技术决策，采用先进的计算工具MATLAB提供的ANN工具箱来建立ANN模型。通过深入分析软基处理问题影响因素，在众多因素中找出了关键因素，构造了两参数决策模型和5参数决策模型，收集ANN模型训练数据，对ANN模型训练试验，结果表明构造的ANN决策模型是有效的。 3.以MATLAB一FIS为工具，构建了高速公路软基处理模糊决策模型。这一工作同样分析了软基处理问题的影响因素，找出了关键因素，将填土高度、软土层厚度和软土的初始孔隙比作为决策模型的输入参数，通过对大量的实际数据的整理统计工作，构造了它们的隶属函数。通过大量的试验，建立了模糊推理规则。对苏嘉杭高速公路江苏段软基处理工程，进行了不同的输入组合试验，结果表明构建的高速公路软基处理模糊决策模型具有实用价值。 4.主要解决了在同一处理层次上(如深层处理)的方法选择问题，本问题的影响因素有些可以定量分析，有些具有模糊性质，只能定性分析。这些因素之间相互关联，相互制约，构成了复杂系统。模糊矩阵理论和层次分析法是系统工程理论中对非定量问题分析的有效决策方法，本研究将二者相结合构造了多层次模糊决策模型。利用此模型分析了武汉市中环线西环段软基处理工程和苏嘉杭高速公路常熟至苏州段软基处理工程，得到了较理想的分析结果。 5.在论述了塑料排水板的排水固结机理基础上，利用自己编写的计算程序，对不同参数的塑料排水板在成层地基、多级等速加载情况下不同时间的固结度，及固结效果进行仿真分析，其结果对塑料排水板的施工参数决策有一定的参考价值。 II 6.近年来遗传算法(GA)已成为解决优化问题的一条新途径，但传统的GA仍存在收敛速度慢，求解效率不高的等缺点，而且对于高维、多约束问题，还会出现收敛困难。为克服传统GA的以上缺点，本文通过引入交叉、变异率自适应调整方法和惩罚策略，藉此提高了算法的求解速率，并首次将改进的遗传算法用于解决深层搅拌桩的最优设计问题，获得满足工程需要、符合设计规范要求且造价最低的设计方案，实现了粉喷桩间距、水泥用量等的优化。 7.提出了基于光纤传感技术的高速公路软基处理施工仿真监控决策的概念，它是是建立在前馈控制的理论基础之上的，通过研制的光纤传感设备和传统的测试方法采集沉降、位移和孔隙水压力等数据，对关键的土工参数进行反分析，对土的强度增长进行估算，然后对下一步的加载进行仿真计算，以确定后续的加载策略。同时能较准确地对土工参数进行估算，较好地解决了工后沉降的估算问题和路面加铺时间的决策问题。更多还原

【Abstract】 Since 1990s,the expressway has developed very quickly in our country. With a rate of about 2 000 km/y, the total mileage of the built expressway has been up to more than 22 500km by the end of 2002, which is the second in the world and exceeds the third one-Canada greatly. However, there are much silt, silty clay, and silt-and-sand bed in the highway grounds of many provinces and cities. Because most of these grounds have high compressibility, great subsidence and bad stability of consolidated drainage, if expressways are built on these grounds without improvement or managed improperly, the quality of roads will fall even destroy. With the building of expressways, the problem of soft ground is becoming more and more protrusive and has been one of key factors influencing project quality, project cycle and project cost. So it has become a very important task to strengthen the research on the improvement of expressway soft ground, which affects the development of traffic project in China.There re many decision problems in the study of expressways ground improvement. During the design phase, decision whether to improve the ground or not should be made firstly, then it should make clearly whether deep improvement or shallow improvement in the segment which needs improvement. And the detailed method on improvement should be made distinctly too. Finally, the technical parameters of the method should be decided. During the construction phase, it should make decision on loading speed according the dynamic supervising results and make decision on the adding spread time of road according the prognostic subsiding result after construction.Owing to various influential factors, a great deal of calculation and imperfect design theory, experience always plays an important part in the technical decision of ground improvement. Therefore, how to make the technical decision of ground improvement more scientific becomes very important. Based on project practice of expressway ground improvement in recent ten years, this article deeply studies many decision problems of expressway ground improvement, makes definitesome existing subjects which should be resolved in the decision of expressway ground improvement currently. Theoretically based on system engineering, artificial intelligence, knowledge engineering, comprehensive evaluate, neural net, fuzzy mathematics, genetic algorithm, decision analysis, computer technology and so on, and combined with the latest developments of geo-technique and soil mechanics, it studies many aspects of problems to establish intelligent decision support system of expressway ground improvement, explores helpfully intelligentization and scientificalization of decision of expressway ground improvement, replenishes the blank of this field.According to the research contents and sequences of chapters and sections, this abstract introduces concisely as follows:1. It expresses briefly the purpose and significance of the article, and introduces comprehensively present research situation, existing problems and research prospect of ground improvement. Based on systems analysis of decision problems of expressway ground improvement, it puts forward research subject and chief contents of intelligent decision support system of expressway ground improvement.2. Artificial neural net (ANN) is used in technical decision of expressway ground improvement for the first time. And ANN model is setting up by usingANN Toolbox which the advanced calculation tool--MATLAB provided. Keyfactor is found out by analyzing deeply the influential factors of ground improvement, and decision models of two parameters and five parameters are set up. By collecting practical data and making experiments on ANN models, the result proves that ANN decision models are valid.3. It uses the MATLAB-FIS for the first time to create the fuzzy decision model on the ground improvement of expressway. It also analyses the influential factors of ground improvement, finds out the key factors, and makes the filling height, the thic更多还原