基于改进支持向量回归的岩溶天窗水位预测模型

Prediction model for water level of sinkholes in karst region based on improved SVM

利用主成分分析对影响变量进行特征提取,选择通用性较强的径向基核函数,应用遗传算法对影响支持向量回归模型的2个重要参数惩罚因子C和核函数参数γ进行优化,建立了基于遗传算法优化的支持向量回归模型,并以提取特征作为模型输入应用于后寨地下河流域平山天窗水位预测.预测结果表明,与传统偏最小二乘回归模型相比,优化后的模型具有更高的精度和更强的泛化能力.

The principal component analysis was employed to extract features of independent variables.The universal RBF kernel function was selected.The genetic algorithm（GA） was applied to the optimization of two important parameters of penalty factor C and kernel parameter γ for the support vector regression model.Accordingly,a GA-based support vector regression model was established.The extracted features of independent variables were regarded as the model input to predict the water level of Pingshan sinkhole in Houzhai underground watershed.The predicted results show that,compared with the traditional PLS model,the proposed coupling model has higher precision and strong generalization ability.