基于粗糙集与支持向量机的混凝土坝施工温度过程预测

Prediction of the Temperature Process for Concrete Dams based on Rough Sets and Support Vector Machine

基于径向基核函数的支持向量机数据挖掘模型,建立大坝施工期温度时程与相关影响因素的关联规则,实现了大坝施工期温度时程快速、实时监测。首先,通过大坝实际浇筑数据的清洗、融合,建立大坝施工期数据库,为模型提供数据快速提取接口。继而,根据粗糙集属性重要度理论进行条件属性约简,剔除冗余属性,确定输入变量;然后,用交叉验证算法,确定模型最优参数;最后,随机选取142个样本作为训练集建立支持向量模型,并使用模型预测剩余的35个样本,模型预测温度时程与实测温度时程基本吻合,模型精度较高且稳定性较好。

A data mining model based on the support vector machine with radial basis functions is developed to determine the association rules between the temperature time history of concrete dams during construction period and relevant factors. Thus,a rapid and real-time monitoring of the temperature process is realized. Firstly,a dam construction information database is established by data cleaning and fusion,which provides a fast data extraction interface for the data mining model. Then,the conditional attribute reduction is carried out to delete the redundant attributes by using the attribute importance theory of rough sets. Finally,optimal parameters of the developed model are determined with the cross validation algorithm. An example with 142 random training samples is established and the remaining 35 samples are predicted. The predicted temperature time history is highly consistent with the measurement,which verifies the accuracy and stability of the model.