中文摘要：

为克服支持向量机对核函数需满足Mercer条件的不足,学者们将相关向量机RVM引入大坝安全监测模型。为进一步提高RVM模型的预测精度,首先通过粒子群算法PSO对RVM的核参数寻优,再利用ARIMA模型对PSO-RVM模型的拟合残差项进行预测修正,建立PSO-RVM-ARIMA模型。通过实例分析,PSO-RVM-ARIMA模型的预测精度和泛化能力较RVM模型均有一定程度的提高。

英文摘要：

To overcome the support vector machine （SVM） for the deficiency of the kernel function to satisfy the Mercer condition , the scholars will introduce vector machines （RVM ） into the dam safety monitoring model .In order to further improve the predictive accuracy of RVM model in this paper ,first of all by particle swarm optimization （PSO） nuclear parameter optimization of RVM ,re‐use of ARIMA model of PSO- RVM model fitting to predict the residual item correction ,PSO- RVM - ARIMA model is estab‐lished .Through the case analysis ,PSO-RVM-ARIMA model prediction accuracy and generalization ability of RVM model has a certain degree of improvement .