基于随机森林算法的大坝应力预测模型的构建及其应用

针对传统统计模型在大坝应力监测时预测精度不高且容易出现过度拟合现象,将随机森林算法引入大坝应力预测中,构建了基于随机森林算法的大坝应力预测模型,对某混凝土重力坝的应力监测数据进行处理、分析和预测,并以平均绝对误差、平均误差平方和及相对误差平方和为指标与多元线性回归模型和神经网络模型进行对比。结果表明,当预测范围在训练集样本范围内时,基于随机森林算法的大坝应力预测模型的预测精度较高,稳定性较好,为大坝应力预测提供了一种新途径。

Development and Application of Dam Stress Prediction Model Based on Random Forest Algorithm

Aiming at the low accuracy and over-fitting phenomenon of traditional statistical models in dam stress monitoring, random forest algorithm was introduced into dam stress prediction. The dam stress prediction model based random forest algorithm was established to process, analyze and predict the stress monitoring data of a concrete gravity dam. The absolute error, the sum of mean square error and the sum of the relative error square were used as indexes to compare with multivariate linear regression model and neural network model. The results show that when the prediction range is within the sample range of the training set, the prediction accuracy of the dam stress prediction model based random forest algorithm is higher and the stability is better, which provides a new way for dam stress prediction.