王晓玲,薛林丽,佟大威,余 佳,祝玉珊,王佳俊
AuthorsHTML:王晓玲，薛林丽，佟大威，余 佳，祝玉珊，王佳俊
AuthorsListE:Wang Xiaoling,Xue Linli,Tong Dawei,Yu Jia,Zhu Yushan,Wang Jiajun
AuthorsHTMLE:Wang Xiaoling，Xue Linli，Tong Dawei，Yu Jia，Zhu Yushan，Wang Jiajun
Unit:天津大学水利工程仿真与安全国家重点实验室，天津 300072
Unit_EngLish:State Key Laboratory of Hydraulic Engineering Simulation and Safety，Tianjin University，Tianjin 300072，China
Abstract_Chinese:灌浆过程中将灌浆功率控制于阈值范围之内，有利于保证灌浆安全和质量．目前施工现场多根据有限个原 位点的灌浆生产性试验，结合专家经验确定灌浆功率阈值．为实现灌浆功率阈值的科学预测，本研究在三维精细裂 隙网络模拟和支持向量回归(SVR)算法两方面提出了改进技术．前者采用改进的拉丁超立方抽样(ILHS)方法模拟出 与实际分布拟合度更高的裂隙参数，从而可构建与岩体实际地质情况一致性更高的三维精细裂隙网络模型，基于建 立的裂隙模型和灌浆实时监控与分析系统采集地质参数和施工参数来构建灌浆功率阈值预测模型的输入参数集，具 体包括：裂隙数量、裂隙平均迹长、裂隙平均倾向、裂隙平均倾角、灌前透水率、孔序、孔深和设计压力．后者采 用改进蝗虫优化算法(IGOA)对 SVR 算法进行改进，实现对惩罚因子 C、核参数 g 以及不敏感损失系数ε的优化计 算，其中 IGOA 中通过引入混沌理论、动态权重和 Lévy 飞行以弥补算法易陷入局部最优的不足，提高算法的搜索 能力．基于 IGOA-SVR 算法构建了灌浆功率阈值预测模型，可实现各灌浆孔孔段的灌浆功率阈值高精度预测．将所 提出的算法和预测模型应用于西南某水电站灌浆工程灌浆功率阈值的预测分析，通过与 4 种常用的预测模型进行对 比，表明其比现有常用模型具有更高的精度．
Abstract_English:In the grouting process，the grout power should be controlled within the threshold to ensure the safety and quality of the grout. Currently，the grout power threshold is determined based on the results of finite in-situ grout tests combined with the experience of experts. To scientifically predict the grout power threshold，in this paper，we propose two improved techniques for three-dimensional fine fracture simulation and a support vector machine(SVR) algorithm. To construct a 3D fine fracture network model that is more consistent with the actual geological condition of the rock mass，the proposed simulation involves improved Latin hypercube sampling(ILHS)to simulate fracture parameters that have a high degree of fitting with the actual situation. Based on the constructed 3D fine fracture net\u0002work model and a real-time grout monitoring and analysis system，geological and construction parameters are estab\u0002lished to construct a set of predictive input parameters for the grout power threshold prediction model. These parame\u0002ters include the number of fractures，the average trace length of fractures，the average dip direction of fractures，the average dip angle of fractures，water permeability before grouting，hole sequence，hole depth，and design pres\u0002sure. Improvement is realized by a proposed improved grasshopper optimization algorithm(IGOA) to optimize the penalty factor C，the kernel function parameter g，and the insensitive loss function ε of the SVR algorithm. TheIGOA is coupled with the chaos theory，dynamic weight，and Lévy flight to prevent it falling into a local optimum and to improve the search ability. Based on the IGOA-SVR algorithm，a grout power threshold prediction model is constructed that achieves high prediction accuracy. As a case study，the proposed algorithm and prediction model are applied to a prediction analysis of the grout power threshold of a hydropower station in Southwest China. A comparison with the results of four prediction models shows that the proposed model has higher accuracy than existing conventional models.
Keyword_Chinese:灌浆功率阈值；三维精细裂隙网络；拉丁超立方抽样方法；蝗虫优化算法；支持向量回归
Keywords_English:threshold of grout power；3D fine fracture network；Latin hypercube sampling；grasshopper optimization algorithm；support vector regression

PDF全文下载地址:http://xbzrb.tju.edu.cn/#/digest?ArticleID=6660