RS-RF模型在混凝土坝变形预测中的应用

Application of RS-RF model in deformation prediction of concrete dam

高性能的混凝土坝变形预测模型作为结构安全性态诊断、预警和科学决策制定的重要参考,同时也是工程效益得以充分发挥的保障措施之一。针对混凝土坝变形监控模型中因子选取的主观性、因子间的多重共线性和预测模型泛化性差等问题,结合粗糙集和随机森林理论在特征属性约简、重要性评价和高精度预测等方面的优势,建立了基于RS-RF的混凝土坝变形预测模型。通过工程实例应用表明:基于RS-RF的混凝土坝变形监控模型可以对影响因子集进行约简并给出各因子的重要性,且预测精度优于常用的SVM模型和RF模型。由此可知,基于RS-RF的混凝土坝变形预测模型实现了影响因子优选,弥补了智能预测模型在定量化分析、预测泛化性等方面的不足,具有较强的工程实用性。

As an important reference for structural safety diagnosis, early warning and scientific decision-making, the deformation prediction model of high-performance concrete dam is also one of the safeguard measures to give full play to the engineering benefits. In this paper, considering the subjectivity of factor selection, multicollinearity among factors and poor generalization of prediction model in concrete dam deformation monitoring model, combined with the advantages of rough set and random forest theory in feature attribute reduction, importance evaluation and high-precision prediction, a concrete dam deformation prediction model based on RS-RF is established. The engineering application shows that the deformation monitoring model of concrete dam based on RS-RF can reduce the set of influencing factors and give the importance of each factor, and SVM model and RF model are commonly used for prediction accuracy. Therefore, the deformation prediction model of concrete dam based on RS-RF has realized the optimization of influencing factors, made up for the deficiency of intelligent prediction model in quantitative analysis and generalization of prediction, and has strong engineering practicability.