摘要
通过利用灌浆过程中的监测数据建立灌浆过程的数学模型,解决机理模型或参数辨识模型不能很好地适应地层和裂隙变化的问题.首先以灌浆管道中流体为对象,依据压力平衡和质量守恒定律来定性分析参数的影响关系,提取孔口压力与注入流量作为反映灌浆过程变化的主要特征参量;然后描述支持向量机建模的基本原理,并研究模型参数优化求取方法,提出嵌套式优化算法求取支持向量,提高建模的速度;最后针对实际工程注入灌浆数据仿真建模.研究结果表明,此方法与实际工程数据吻合较好,本次建模统计出错率不超过为7%.
A novel grouting process modeling method based on support vector machine（SVM） tool was proposed,which could avoid the difficulty of the common mathematical mechanism model for the invisible changes of stratum and varied rock facture width.Firstly a few sub-mathematical mechanism models were established based on pressure equilibrium and mass conservation.Through analyzing the mechanism model,pressure and filling flow of grouts could be described as the change of rock facture and stratum,so they were chosen as the characters of SVM.Then studied the principle of structural risk minimization of SVM was studied,an embedded uniformity optimizing algorithm was introduced to selection of SVM model parameters.The embedded uniform algorithm is superior to grid optimization algorithm in arithmetic speed,which is easier for project applied requirement.At last,the model of grouting process is simulated based on SVM method.The results show the new SVM method is adaptive with the data of grouting project,the error of modeling is lower than 7%.
出处
《长沙理工大学学报（自然科学版）》
CAS
2011年第1期51-55,共5页
Journal of Changsha University of Science and Technology:Natural Science
基金
湖南省教育厅科研资助项目(50875028)
关键词
支持向量机
参数优化
灌浆
非线性过程
智能建模
SVM
parameter optimization
grouting
nonlinear process
intelligent modeling
