【作者】 胡珉；

【导师】 吴耿锋；

【作者基本信息】 上海大学 ， 控制理论与控制工程， 2006， 博士

【摘要】 免疫理论研究的深入和智能技术的发展，为产生新的智能算法提供了良好的研究基础；而随着地下空间的开发新技术不断出现带来的新的挑战，又为智能计算提供了广阔的应用舞台。本文借鉴免疫系统中MHC(主要组织相容复合体)调控原理，从工程需求出发，进行了多角度的深入研究，提出了相关的智能算法，成功地在盾构法隧道工程的实际施工中得到应用。本文回顾了有关免疫系统的主要研究成果以及人工免疫系统理论的国内外研究现状，深入理解了MHC的免疫调控原理。在分析了目前盾构法隧道施工中主要遇到的问题与原因以后，提出了学术研究与工程研究相互结合和渗透的新设计理念，开展了组合优化、数值优化、公式发现和自动控制等多方面的研究。第二章基于MHC免疫调控原理，提出了一个基于MHC调控的免疫组合优化的新算法IOAMHC。而第三章将MHC调控的思想应用到了数值优化算法的设计中，并结合均匀设计理念，提出了一种基于MHC调节的免疫进化新算法MHCIEA。这两个算法利用了MHC的自调节特性，指导抗体进化。与其他的优化算法相比，由于优化搜索思想不同，优化速度得到了提高，优化能力得到了加强。通过对基准问题的计算，充分验证了算法在性能上的优势，同时算法也通过了实际隧道施工的检验，取得了良好的指导作用。第四章提出了基于MHC调控公式发现算法IFDA，扩大了免疫算法的应用领域，为免疫算法的发展提供了新的途径。IFDA成功地运用在双圆盾构地面沉降的横向和纵向公式发现方面，通过与历史的工程数据比对，证明通过该算法得到的地面沉降公式对工程数据具有很好的拟合度，充分反映了双圆盾构的地面沉降规律。第五章针对隧道和软土地基多种物理对象和施工过程控制问题，借鉴生物免疫系统面临不确定的外来抗原呈现出的超强识别能力，创造性地提出了一种基于免疫系统模型的多模型控制新算法IMMC。该算法解决了非线性、不确定复杂系统控制困难的问题。通过计算机仿真实验，表明算法在系统发生突变时具有良好的适应性能。同时，该算法在双圆盾构法隧道施工对地面沉降控制中，取得了良好的控制效果。IMMC算法是生物学理论、控制理论以及计算机技术成果多学科交叉应用的成果。在本文研究成果的指导下完成的“盾构法隧道远程智能系统”和“盾构法隧道施工管片(错缝)拼装选型智能系统”被应用于上海地铁和越江隧道等近三十项工程，取得了显著的社会效益、环境效益和经济效益。更多还原

【Abstract】 The development of biological immune system theory and intelligence technology brings new ideas and great opportunities for the generation of intelligence algorithms. Also the new requirements of exploring underground space bring the challenge and broad arena for the application of intelligence algorithms. Inspired by MHC (Major Histocompatibility Complex) regulation principle of immune theory and its research in various fields, this dissertation proposes a series of novel and innovative intelligence algorithms and successfully applies them to the engineering practices for the tunnel engineering.The dissertation firstly reviews the main research results on immune system, then analyzes the current situation for both the domestic and foreign research in the artificial immune system and goes deep into the studies of MHC regulation principle. After having summarized & understood the main difficulties and their root cause in tunnel construction, the dissertation puts forward the new ideas on mutual incorporation and penetration of academic research and engineering study. These researches focus on the combination optimization, numerical optimization, function discovering and automatic control.The second chapter of the dissertation proposes an immune combination optimization algorithm (IOAMHC) based on MHC regulation. The third chapter describes the application of MHC regulation principle to numerical optimization algorithm and proposes a new immune numerical optimization algorithm (MHCIEA) combined with the concept of uniform design. Both algorithms are using the character of MHC self-adjust to guide the antibody evolution. Compared with the other optimization algorithm, the performances of the IOAMHC and MHCIEA are obviously improved because of the enhanced optimization searching strategy. The experiment for the benchmark cases shows that the performances of the algorithms are outstanding, and the application in the tunnel engineering also confirmed the good effect of new algorithms.The fourth chapter proposes an innovative formula discovering algorithm (IFDA) based on MHC regulation, which enlarges the research fields of immune algorithm and provides the new development space for immune algorithm. IFDA is applied to self-searching formula of the horizontal and vertical settlement curve for DOT (Double-0 Tube) shield tunneling. Compare with the historical engineering data, the formula is proved to fit the actual data closely and reflect the rule of ground settlement successfully.In the fifth chapter, a new method based on immune system for multiple model control (IMMC) is introduced. The method simulates the strong ability of discrimination to the uncertain antigen in the biological immune system and tries to solve the difficulties in controlling the nonlinear and uncertain complicated system effectively and rapidly. The simulation experiment shows that the algorithm has excellent performance as it can rapidly adapt to the sudden changes in the control system. The algorithm is successfully applied to the settlement control in a DOT shield tunneling. Actually IMMC is an interdisciplinary algorithm which covers biological theory, control theory and computer science.Directed by the research results of the dissertation, two systems named "Shield Tunneling Remote Intelligent System" and "Shield Tunneling Segment Type Selection System" are successfully implemented. Both systems led to great benefits toward society, environment and economy as they are being used in about thirties engineering projects, such as Shanghai metro, cross-river tunnel etc.更多还原