基于BP神经网络的公路工程投资估算预测模型研究
(甘肃省交通规划勘察设计院股份有限公司,甘肃 兰州 730010)
- 文献要素
-
摘要:公路工程投资估算传统编制方法主要是依托设计工程量套用估算指标进行编制,存在耗时费工、准确率低、受人为因素影响大等问题。为了在投资决策阶段将历史项目施工图预算数据用于新建项目投资估算,进而对传统方法编制的估算进行验证,文章提出一种基于BP神经网络的公路工程投资估算预测模型。首先统计50个已完工公路项目的施工图预算数据,采用专家打分法,对不同影响因素赋权,依据权重高低依次选取了地域、路线长度、桥隧比等权重较高的16个主要因素作为BP神经网络的输入层,再将项目的建安费和总造价作为输出层,采用Python程序中的NumPy库建立预测模型。最后将统计样本的80%作为训练样本对预测模型进行训练,将剩下的20%作为测试样本对模型的准确性进行测试,发现误差控制在±10%以内,符合投资估算的精度要求。关键词:公路工程;投资估算;BP神经网络;快速预测Abstract:Traditional methods for highway engineering investment estimation primarily rely on applying estimation metrics to designed quantities,which often leads to time-consuming processes,low accuracy,and high susceptibility to human factors. To utilize historical working drawing budget data from previous projects for new project investment estimation during the decision-making stage,and thereby validate estimates derived from traditional methods,this paper proposes a prediction model for highway engineering investment estimation based on a BP neural network. The study compiled working drawing budget data from 50 completed highway projects,used the expert scoring method to assign weights to various influencing factors,and selected 16 major factors—such as region,route length,and bridge-tunnel ratio—based on their weights as the input layer of the BP neural network. The construction and installation costs along with the total project cost were set as the output layer. The prediction model was built using the NumPy library in Python. Finally,80% of the sample data was used to train the model,while the remaining 20% was used to test its accuracy. The results showed that the error was controlled within ±10%,meeting the precision requirements for investment estimation.Keywords:highway engineering;investment estimation;BP neural network;predict quickly参考文献[1] Martin T.Hagan,Howard B.Demuth. Neural network design[M].成都:机械工业出版社,2018.[2] 陈明.MATLAB神经网络原理与实例精解[M].北京:清华大学出版社,2013.[3] 交通运输部.JTG/T 3821-2018公路工程估算指标[S].北京:人民交通出版社,2018.[4] 马永军,杨志远.基于模糊神经网络的公路造价估算模型探究[J].公路工程,2017(6):41-47.[5] 汪优.基于BP神经网络的人工单价动态预测模型研究[J].工程管理学报,2021(2):39-43.[6] 李芬,汤虎.基于神经网络的山区公路桥梁造价预测[J].土木工程与管理学报,2011(4):78-81.[7] 李俊达,李远富.基于CBR的公路工程造价估算模型[J].公路交通科技,2020(6):44-49.[8] 刘仁杰.基于模糊BP神经网络的建筑工程造价模型[D].济南:山东建筑大学,2023.[9] 谢洵.基于BP神经网络的高速公路工程造价估算模型研究[D].成都:西南财经大学,2022.[10] 李铭康.基于机器学习的工程造价预测算法对比研究[D].广州:华南理工大学,2022.[11] 李驰宇.高速公路造价快速估算模型与方法的研究[D].成都:西南交通大学,2006.建筑经济,2025(10):48-52
