摘要
鉴于大坝安全监测评价中相关向量机(RVM)模型性能的优劣取决于核函数选择的问题,采用混合核函数(即结合局部核函数和全局核函数)进一步提升RVM模型在大坝安全建模过程中的拟合预报精度,并采用改进粒子群算法对其寻优。经实际工程验证,基于混合核RVM理论的优化模型可在一定程度上提升模型的性能。
Kernel function has important influence on the performance of the RVM model for evaluation of dam safety monitoring. The mixed kernel function which is composed of local and gIobal kernel kernels is used to increase the fitting accuracy of model for dam safety modeling. The optimization parameters of RVM model are sought by an improved parti- cle swarm algorithm. The results of instances show that the hybrid kernel RVM optimization model can enhance predic- tion accuracy to some extent.
出处
《水电能源科学》
北大核心
2015年第5期59-61,89,共4页
Water Resources and Power
基金
重庆市饮用水源地(水库)水资源涵养及水生态修复植物配置规划项目(渝水科201306)
