【作者】 陈磊；

【导师】 仲景冰；

【作者基本信息】 华中科技大学 ， 建筑与土木工程， 2015， 硕士

【摘要】 我国目前基础建设工程快速发展,大量工程建设都与基坑密切相关。基坑工程具有复杂性和不可预计性的特点,研究基坑围护结构的侧移预测能为基坑的安全性和稳定性提供参考,促进对基坑的设计和施工进行优化,有利于预防和减少基坑相关事故,降低人员伤亡和财产损失。本文研究分析了目前国内外基坑围护结构变形的智能算法,比较并总结这些算法运用在基坑工程中的优势和局限性;引出遗传规划理论并分析其原理和特点,在此基础上提出运用遗传规划模型预测基坑围护结构的变形。本文还介绍了基坑围护结构变形机理和类型,着重探讨开挖深度、开挖阶段、平面尺寸、支撑刚度、围护长度和厚度等影响围护结构侧向位移的因素。重点研究遗传规划模型的建立和各参数设置,主要包括适应度函数、选择概率、交叉概率、变异概率等。同时利用工程数据对模型进行训练和测试,筛选出最终结果表达式,与神经网络模型所得结果进行对比,分析两者差异性,检验本表达式的精度要求。最后利用武汉地铁深基坑实际开挖数据对最终结果进一步验证,证明本模型在基坑围护结构侧移预测方面的可靠性和实用性。本文的研究涉及基坑工程领域和计算机编程领域,使用新的角度将不同的领域有机结合起来,为深基坑围护结构的侧向变形预测提供了新的解决思路,推动了遗传规划在基坑领域的实际工程运用,并具有较高运用价值。同时对于基坑安全和创造经济效益更多价值具有重要的现实意义,能够为基坑工程变形和为类似工程提供有益的参考。更多还原

【Abstract】 Infrastructure construction projects are developing rapidly in China, a large number of construction projects are closely related to deep foundation pit. Foundation pit engineering is complex and unpredictable, so the research for lateral displacement prediction of foundation pit bracing structures can provide the reference for the foundation safety and stability, promote the optimization of design and construction for foundation pit, and it can be conducive to the prevention and reduction of the foundation pit accident, reducing casualties and property losses.This thesis was analyzed the present domestic and foreign intelligent algorithm of the deformation of the bracing structure of foundation pit, compared and summarized their advantages and limitations when applied in foundation pit engineering. Then it was introduced genetic programming model and the analysis of the model principle and characteristics. Base on the theory, genetic programming model was used to forecast the deformation of foundation pit. The deformation mechanism and types foundation pit was introduced and it focused on the displacement influence factors such as depth of excavation, dimension, strut stiffness, wall thickness and length and excavation stage. The key work was the establishment of genetic programming model and the parameters set, including the fitness function, the probability of selection, crossover and mutation probability. Meanwhile, the engineering data was used to train and test the model, screening out the final result expression, and then the results were compared with the results of the neural network model to check the accuracy and reliability. In the end, Wuhan metro deep foundation pit excavation data was used to validate the final result, the results proved that the model for displacement prediction of the bracing structure of foundation pit was reliable and practical.This study involves fields of excavation engineering and computer programming, combining different areas in a new angle. It provides a new idea for the lateral deformation prediction of deep foundation promoting practical engineering application of genetic programming. The thesis also provides important practical significance for excavation safety and economic value, which is useful for excavation deformation and providing useful reference for similar projects.更多还原