基于改进随机森林算法的渡槽位移及应力预测模型

渡槽位移及应力监测是渡槽健康监测的重要内容,其位移及应力变化与多种因素存在复杂的非线性关系,这种关系导致传统数学模型难以较准确地预测出不同环境变量影响下渡槽的位移及应力变化情况。对此,提出了一种基于随机游走思想的随机森林算法,该算法以渡槽水位、气温及水温为输入,能较准确地预测出渡槽不同测点的位移及应力。最后,通过一个数值算例,对比了该算法与已有模型算法的拟合性能和泛化能力,验证了该算法的优越性,可满足渡槽工程位移及应力预测的需要。

Aqueduct Displacement and Stress Prediction Model Based on Improved Random Forest Algorithm

Aqueduct displacement and stress monitoring is an important part of aqueduct health monitoring. Its displacement and stress changes have a complex nonlinear relationship with various factors. This relationship makes it difficult for the traditional mathematical models to accurately predict the aqueduct displacement and stress change under the impact of different environmental variables. Therefore, this paper proposes an aqueduct displacement and stress prediction model based on improved random forest algorithm. The model uses aqueduct water level, air temperature and water temperature as input, and can accurately predict the aqueduct displacement and stress at different measuring points. Finally, the fitting performance and generalization ability of the method and the existing model algorithm are compared through a numerical example. The superiority of the method is verified, which can meet the needs of aqueduct engineering displacement and stress prediction.