【作者】 肖清华；

【导师】 张继春；

【作者基本信息】 西南交通大学 ， 岩土工程， 2007， 博士

【摘要】 爆破技术以其灵活、方便、快速的特点在我国大部分隧道修建中得到广泛应用。隧道掘进爆破设计的质量、速度直接影响到隧道修建的安全、质量、成本和工期，如何改变目前设计人员凭经验确定设计方案、选取单循环进尺及爆破参数、手工计算和手工绘图、计算不精确、调整修改困难、设计速度慢以及人工编制施工进度和进行施工质量管理的落后局面，已成为爆破工程、隧道工程界亟待解决的问题之一。因此，研制出功能完善的隧道爆破设计智能系统，不仅具有重要的工程应用价值，而且在提高我国爆破设计水平、设计质量和设计速度的同时，更推动了岩石爆破理论与计算机技术和人工智能的紧密结合。本文针对目前隧道爆破开挖施工中的方案确定、爆破参数选取、炮孔布置与绘图、计算速度慢、数据管理困难、难以借鉴以往爆破资料指导后续设计等科学问题进行研究。综合运用隧道掘进爆破设计原理、计算机数据库技术、人工智能理论、神经网络技术、数学建模理论和计算机图形学原理，实现了用计算机方法替代人工进行隧道爆破设计的目标，研制出隧道爆破设计智能系统，以促进隧道爆破设计的智能化进程。按照隧道掘进爆破的设计原理和方法提取与爆破设计相关的参数和数据及其相关的专家经验知识信息，依据数据库设计技术方法并结合提取的参数和数据，确定了适合隧道爆破设计数据的数据库及其数据表结构。由此，选择Microsoft Access系统创建了隧道爆破设计数据库bps，并根据数据表结构创建了全部数据表及其主键、索引和关系，使得隧道爆破数据库在操作和管理上达到了数据的安全性、一致性和完整性的统一。根据传统的隧道爆破设计技术并结合计算机智能系统理论，按照软件工程方法研究提出隧道爆破设计智能系统由知识库、数据库、推理机、人机交互系统和解释机构等五部分组成，并给出了基于人工智能的一系列推理机制和搜索策略。在综合考虑安全、实用、质量、用户特点等多种因素的基础上，创建了系统管理、参数智能计算、爆破数据、布孔设计、施工设计和施工信息管理等系统功能模块。结合目前各类隧道断面轮廓线的特点，提取出其断面参数和特征参数，建立了可用于计算机编程的隧道断面轮廓线数学模型，通过给出的隧道断面参数输入方法及算法步骤与流程，可实现任意隧道断面轮廓线的精确绘制，并给出了准确计算断面面积的计算步骤和计算公式，为隧道爆破的炮孔自适应布置奠定了基础。根据隧道爆破设计原理和计算机技术要求，结合数学建模理论，提出了掏槽、周边、辅助、掘进等炮孔的自适应布置原则与方法，建立了相应的数学模型。该研究工作为系统实现提供了理论支持，系统据此可实现按隧道爆区地质条件和断面几何尺寸自动搜索选取最优爆破参数，并以此在隧道断面上进行自适应布置各类炮孔，且可按给出的布孔修改原则和方法进行自动调整。运用人工神经网络理论与方法，结合隧道爆破设计的实际需要，以岩石f值、隧道断面面积、实际进尺、炮孔直径、平均线性超挖、平均线性欠挖为网络输入参数，以设计进尺、炸药单耗、周边孔距、辅助孔孔距、辅助孔排距、掘进孔孔距、掘进孔排距为网络输出参数，建立输入层节点数为6、输出层节点数为7、隐含层节点数为3的三层结构隧道爆破设计BP网络，并提出基于爆破先验知识的神经网络模型学习约束条件，实现在借鉴原有爆破设计参数和爆破效果的基础上快速、准确计算爆破参数的目标。选择Visual C++作为隧道爆破设计智能系统的开发平台，按系统各个功能模块及其实现方法编制程序，使系统框架下的各个相互独立的模块既能顺利完成各自功能又能成为一个相互关联并运转良好的有机整体。将开发完成的隧道爆破设计智能系统(ZXTBS)应用于武广铁路客运专线上连溪和乐善亭隧道爆破施工，设计出循环进尺分别为2.2m、2.3m和2.8m三种爆破方案。7次爆破施工的实践表明，隧道爆破设计智能系统给出的各种设计图表可直接应用于工程施工，能实现爆破参数优化、计算机自适应布孔成图和系统自我学习等目标，并在20分钟的时间内完成隧道爆破设计；爆破质量达到工程要求，并解决了施工中隧道掌子面不平整、炮孔利用率低、雷管脚线易炸断、超欠挖等问题，提高了施工质量和进度，节省了施工费用。更多还原

【Abstract】 Because of the characteristics of agility, convenience and speediness of blasting technique, it is applied in most of tunnel construction in our country. The quality and velocity of the tunnel blasting design can directly influence the safety, quantity, cost and work period. How to change the current predicaments that the design determined, blasting parameters selected and calculated by the designer’s experience, handwork drawing, adjusting and also those problems, such as, modifying difficulty, slow velocity on designing, deficiency on construction schedule authorized by people and construction quality control would be a very urgent problem to solve in the tunnel blasting and tunnel engineering. Therefore, to develop an intelligent tunnel blasting design system not only has important value in application function, but also will improve blasting design level, quality and velocity, even the theory of rock blasting and the combination between computer techniques and artificial intelligences.This thesis aims and argues at the scientific problems of the design determining, blasting parameters selecting, blast-hole layout, slow calculating, difficulty of data management and to utilize the former designs by using the existed blasting data, and continuing design as well. The target of replacing manual by the computer to complete the tunnel blasting design can be implemented by using tunnel blasting theories, computer database techniques, AI theories, ANN techniques, mathematics model theories and computer graphics theories. So, the intelligent process of the tunnel blasting design would be accelerated.According to the tunnel blasting design theories and methods, the parameters, data and expert experience knowledge related to the blasting design were obtained. The database structure and its datasheet structure suited to the tunnel blasting design were determined by adopting database techniques and methods. Selecting the software of Microsoft Access as the database system, the tunnel blasting database bps was set up. The all datasheet and their main keys, indexes and connections each other were established, so that the tunnel blasting design and construction management would be more security, consistency, integrality.According to the regular tunnel blasting design techniques, the composition of intelligent system for tunnel blasting design was determined by adopting the methods of both software engineering and artificial intelligence, which is consisted of knowledge base, database, reasoning system, computer interaction system, and explanation organization, on which a series of reasoning system and search policies based on the AI were given. The function modules including the system management, intelligent calculation of parameter, blasting data, primary design, construction design and construction information management are established on the basis of multi-factors with safety, practicality, quality and users’ characteristic.The tunnel cross-section parameters were picked up according to the characteristics of the tunnel outline. The mathematical model, the tunnel outline could be drawn automatically and accurately by the method of input parameters of tunnel, the algorithm process and data flow given. Also the formulas and calculating steps of the tunnel cross-section area calculation could be drawn, which is also the foundation to self-adapting blast-hole disposal for the tunnel blasting.According to the computer techniques and tunnel design theories, together with mathematics modeling theories, the mathematics models were created, which can provide theory sustaining. The system can be self-adapting blast-hole disposal, in the aspects of the contour-hole, reliever-hole and excavated-hole by the tunnel geologic conditions and cross-section sizes. On the basis of above all, the blast-hole arrangement can be modified and adjusted.Using ANN theories and combining the practical requirement of tunnel blasting, a three-layer BP neural network model with 6 nodes in input layer, 7 nodes in output layer, 3 nodes in connotative layer is constructed, in which the Protodiakonov’s hardness coefficient, tunnel cross-section area, practical advance per round, blast-hole diameter, etc. are considered as the input parameters of network, and the design advance per round, powder factor, contour-hole spacing, relief-hole spacing, excavated-hole spacing as the output parameters of network. The restrain conditions of network study, on the basis of the blasting prior knowledge according to the relation between the input and output parameters of the network, are proposed to accelerate calculating. The neural network model can implement the goal that the parameters of tunnel blasting are calculated accurately and quickly by using the existed blasting data quantitatively.Taking Visual C++ as the tool for the application system programming, it is available for the tunnel blasting design intelligent system to transfer the parameters, principle of layout blast-hole and relative mathematics models into system function, thus it could become an organic self-governed organization with all the functions operating well inside the system frame.The tunnel blasting design intelligent system(ZXTBS) was applied to Shang-lianxi and Le-shanting tunnel of Wuhan-Guangzhou passenger transport special railway. Considering the facts of the tunnel, the designs with 2.2m, 2.3m and 2.8m of the per ground were put forward separately, which achieved good blasting effect. It is shown that the design diagrams provided by the ZXTBS system can be used to engineering construction and it can achieve good implement on the goal that blasting parameters optimizing, computer drawing of self-adapting layout blast-hole, and self-studying. The designers can complete the tunnel blasting designs accurately and quickly by using ZXTBS system within 20 minutes. A series of problems such as the section was not smooth, the blast-hole efficiency was low, and detonator was fragile, etc., were solved; it can improve the construction quality, promote the schedule and reduce the construction cost.更多还原