基于EMD-ANN耦合模型的路基沉降预测方法研究  

Research on Roadbed Settlement Prediction Method Based on Improved Artificial Neural Network Modeling

作　　者：潘文超 PAN Wenchao(China Railway SIYUAN Survey and Design Group Co.,Ltd,Wuhan 430063)

机构地区：[1]中铁第四勘察设计院集团有限公司交通院,武汉430063

出　　处：《铁道勘测与设计》2024年第4期47-51,共5页Railway Survey and Design

摘　　要：本研究基于经验模态分解(EMD)与人工神经网络(Artificial Neural Network,ANN)的耦合算法,致力于路基沉降预测方法的深入研究。在道路工程中,路基沉降是一个关键问题,直接影响着道路的安全性和可持续性。为了提高沉降预测的准确性和可靠性,本研究应用EMD将复杂的路基沉降时程信号分解成多个本征模态函数,有效提取信号的局部特征。通过分析各个本征模态函数的特征,确定最具代表性的模态函数为序列趋势项,运用ANN模型对各本征模态函数进行训练,以捕捉各个模态函数之间的非线性关系。通过模型的训练和优化,得到高度适应路基沉降特性的改进ANN预测模型。将经过训练的ANN模型用于对广州某路基沉降进行预测,实验结果表明,采用改进ANN算法的路基沉降预测方法在准确性取得了显著的提升。与传统方法相比,本研究提出的方法能够更好地适应路基沉降信号的复杂变化,并具有更强的泛化能力。因此,该研究为道路工程领域提供了一种先进而可行的沉降预测方法,为提高路基设计的可持续性和安全性提供了有力支持。This study is dedicated to the in-depth research of roadbed settlement prediction methods based on the coupled algorithm of Empirical Modal Decomposition(EMD)and Artificial Neural Network(ANN).In road engineering,roadbed settlement is a key issue,which directly affects the safety and sustainability of roads.In order to improve the accuracy and reliability of settlement prediction,this study applies EMD to decompose the complex roadbed settlement time-range signal into multiple eigenmode functions,and effectively extract the local features of the signal.By analyzing the characteristics of each intrinsic modal function,the most representative modal function is identified as the sequence trend term,and the ANN model is used to train each intrinsic modal function to capture the nonlinear relationship between each modal function.Through the training and optimization of the model,an improved ANN prediction model that is highly adapted to the settlement characteristics of the roadbed is obtained.The trained ANN model is used to predict the settlement of a roadbed in Guangzhou.The experimental results show that the accuracy of the roadbed settlement prediction method using the improved ANN algorithm is significantly improved.Compared with the traditional method,the method proposed in this study can better adapt to the complex changes of roadbed settlement signals and has stronger generalization ability.Therefore,this study provides an advanced and feasible settlement prediction method in the field of road engineering,which provides strong support for improving the sustainability and safety of roadbed design.

关 键 词：路基沉降预测 改进人工神经网络 经验模态分解 道路安全 本征模态函数 

分 类 号：U41[交通运输工程—道路与铁道工程]