基于APPSO-RVM与APPSO-SVM的大坝安全预警模型的应用比较研究

Application and Comparison of SVM and RVM Algorithm in Dam Safety Modeling

为了提高大坝安全预警模型的精度和泛化能力,基于支持向量机(SVM)和相关向量机(RVM)理论,利用自适应粒子群算法(APPSO)分别对SVM和RVM模型中的参数进行寻优,建立了基于APPSO-SVM与APPSO-RVM的大坝安全预警模型,并通过实例应用做了比较。结果表明,尽管APPSO-RVM模型的相关向量个数少于APPSO-SVM模型,但APPSO-RVM模型的拟合精度和泛化能力均高于APPSO-SVM模型,因此在实际建模时应优先选择该模型。

Based on support vector machine（SVM）and relevance vector machine（RVM）theory,the parameters are optimized through adaptive particle swarm algorithm（APPSO）to improve fitting accuracy and generalization ability of dam safety warning model.So,we set up dam safety warning model based APPSO-SVM and APPSO-RVM.Compared SVM with RVM,example＇s results indicate that although the number of relevance vector of APPSO-RVM model is less than APPSO-SVM model,the fitting accuracy and generalization ability of APPSO-RVM is better than APPSO-SVM model.Therefore,we should choose APPSO-RVM model in reality.