A factor mining model with optimized random forest for concrete dam deformation monitoring
Hao Gu, Meng Yang, Chongshi Gu, Xiaofei Huang
Abstract
The unique structure of a dam complicates safety monitoring. Deformation can provide important information about dam evolution. In contrast to model prediction, actual dam response monitoring data can be used for diagnosis and early warning. Given the poor data mining ability of the conventional methods, it is essential to develop a method for extracting the factors influencing a dam. In this study, a data mining method and a model for evaluation of concrete dam deformation were developed using the evidence theory and a random forest. The model has the advantages of being easily understood, visualization with low complexity of training time, and accurate prediction. The model was applied to an actual concrete dam. The results indicated that the proposed random forest model could be used in analysis of concrete dams.