摘要
针对大坝变形具有强非线性的特点以及在采用传统神经网络模型进行预测时存在局部极小、过学习等问题,提出一种新的大坝变形预测方法——支持向量机方法。该方法基于统计学习理论,采用结构风险最小化原则,保证了模型具有很强的泛化性能,并通过求解一个二次规划问题确保模型具有全局最优。以东江大坝变形预测为实例,说明了该方法的可行性和有效性。
Because of the strong non-linear property of dam deformation, and the existence of local minima, overlearning of conventional neural networks model, a new model of dam deformation prediction is presented based on Support Vector Machine （SVM）. The model is based on statistical learning theory and structure risk minimi- zation principle. It serves good generalization ability. By solving a quadratic formulation problem, a global optimum can be found. The feasibility and validity of SVM deformation prediction model are showed by the practical sample of deformation prediction of Dongiiang dam.
出处
《测绘工程》
CSCD
2007年第6期1-4,共4页
Engineering of Surveying and Mapping
基金
国家自然科学基金资助项目(40474003)
关键词
大坝变形预测
支持向量机
神经网络模型
dam deformation prediction
support vector machine
neural networks model
