RBF神经网络—半参数优化模型在滑坡预报中的应用  

The Application of Optimized RBF Neural Network Semi-Parametric Model in Landslide Forecasting

作　　者：易锋[1] 胡家赋[1] 郭军[1] 

机构地区：[1]广州海洋地质调查局,广东广州510760

出　　处：《测绘与空间地理信息》2014年第12期43-45,共3页Geomatics & Spatial Information Technology

基　　金：国土资源部海底矿产资源重点实验室开放基金(1212011220113)资助

摘　　要：将RBF神经网络模型与基于补偿最小二乘准则的半参数模型相结合,首先进行RBF神经网络模型的预报,在此基础上进行半参数模型改正,非参数利用补偿最小二乘法求出,建立优化的RBF神经网络——半参数模型。结合某边坡的GPS高程观测数据进行建模预测,与单一采用RBF神经网络预报模型相比,结果显示精度较好,该方法有效可行。This paper combined RBF neural network model with serialized least squares criterion based on the semi - parametric model to establish optimized RBF neural network semi - parametric model, firstly we forecasted by RBF neural network model, then we cor- rected the result in semi -par~,metric model , and the non -parametric was derived by serialized least squares. Using this method to model and forecast a slope of GPS observational data, compared with RBF neural network model, the results show good accuracy to prove this method effective and feasible.

关 键 词：半参数模型 RBF神经网络 滑坡预报 

分 类 号：P25[天文地球—测绘科学与技术] TP183[自动化与计算机技术—控制理论与控制工程]