欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
在新建水利枢纽、大坝除险加固以及监测系统不稳定等原因导致有效观测序列较短的情况下，为了对大坝变形进行有效的监控和预测，用基于不敏感损失函数的支持向量回归机（ε-SVR）对大坝变形进行数值模拟。采用支持向量回归机、序贯极小优化算法对大坝变形水平位移各分量的特征因子进行选择，模型建立后分离水压、温度、时效等主要分量，并分析其对变形的影响。对实际工程进行数值模拟的结果表明，依靠较少的支持向量就可以建立模型，该模型具有拟合效果稳定、预测结果准确可靠的特点，可以应用于大坝短期观测资料的分析。
英文摘要：
Under the condition of the short effective observation sequence which is related to the short operation period of new hydro-junction, dam danger control and reinforcement as well as the instability of dam monitoring system, and for the effective monitoring and forecasting of dam deformation, the method of support vector regression based on insensitive loss function （ε- SVR） is used for the numerical stimulation of dana deformation. The support vector regression and the sequential minimal optimization algorithm are used for the characteristic factors selection of the dam horizontal displacement. The main components, such as water pressure, temperature and aging are divided and their effects on dam deformation are analyzed after the simulation model is set up. The numerical simulation method has been used for practical construction, and the simulation results show that the model can be built with a small quantity of support vectors and the method is stable in fitting and accurate in forecasting. So the ε-SVR model can be widely used for short-term data analysis for dams.
同期刊论文项目
同项目期刊论文
Structure analysis for concrete-faced rockfill dams based on information entropy theory and finite e
Integrated parameter inversion analysis method of a CFRD based on multi-output support vector machin
位置：