铁道标准设计2025，Vol.69Issue(4)：93-101,9.DOI:10.13238/j.issn.1004-2954.202309070001
融合多维空间相似理论与机器学习算法的山区铁路桥梁智能选型方法
Research on an Intelligent Method for Railway Bridge Selection in Mountainous Areas Integrating Multi-Dimensional Spatial Similarity Theory and Machine Learning Algorithms
摘要
Abstract
To fully explore existing bridge case knowledge and expert experience,simulate professional engineers in bridge selection,and improve the level and efficiency of bridge selection,this study proposes an intelligent method for railway bridge selection in mountainous areas by integrating multi-dimensional spatial similarity theory and machine learning algorithms.By comprehensively considering three levels of engineering data,topography and geology,and hydrometeorology,11 feature attributes were selected to establish a hierarchical index system for the selection of railway bridges in mountainous areas,with the weight of each index determined using a combination weighting method.Based on historical bridge case data,the cases were characterized following specific entry rules and case characterization methods.GIS was used to divide attribute units of existing cases and design entry rules.By integrating multi-dimensional spatial similarity theory,a GIS case database for railway bridges in mountainous areas was established.The similarity between target cases and existing cases was calculated,and a similarity criterion was designed for case retrieval and suggested applications.The nearest neighbor retrieval strategy was incorporated into the case retrieval process,and the existing cases obtained from the retrieval were used as the sample dataset.A BP neural network model was designed to simulate the training and learning of the human brain for intelligent selection of railway bridge types.Additionally,three regression prediction methods—decision tree,K-nearest neighbor,and support vector machine—were used for intelligent railway bridge selection,with the selection results compared with the prediction accuracy of the BP neural network algorithm.A railway bridge in a mountainous area was used as an engineering case to validate the method.The results showed that the model's selection results were consistent with the actual bridge type.This method is applicable to the selection of railway bridges in the challenging mountainous areas of western China,while providing new insights for intelligent route selection design.关键词
多维空间相似理论/机器学习算法/桥梁智能选型/GIS案例库/相似性判别准则/组合赋权法Key words
multi-dimensional spatial similarity theory/machine learning algorithm/intelligent bridge type selection/GIS case database/similarity criteria/combination weighting approach分类
交通工程引用本文复制引用
李媛,韩峰,白如博,魏昊..融合多维空间相似理论与机器学习算法的山区铁路桥梁智能选型方法[J].铁道标准设计,2025,69(4):93-101,9.基金项目
国家自然科学基金项目(51568037) （51568037）
