首页|基于PCA-BP神经网络的既有建筑改造成本预测

基于PCA-BP神经网络的既有建筑改造成本预测

扫码查看
既有建筑由于建造过程信息的缺失、损毁,缺乏造价定额资料等成本估算标准数据,导致决策阶段很难快速准确地计算出改造工程造价.针对该问题,提出了基于主成分分析(PCA)和BP神经网络的既有建筑改造成本预测的方法.通过案例与文献分析,识别并提炼出影响既有建筑改造成本的10个影响因子.利用主成分分析方法,对提取的10个因子进行降维,归纳出3个新的综合变量.采用BP神经网络对280个既有建筑改造成本进行分组训练、验证与测试.结果显示,PCA-BP神经网络模型基于降维且消除指标之间相关性数据为基础,提高了预测效率,方根误差、相关度均取得了较好的结果,实测数据与预测数据之间的综合误差为2.66%,为既有建筑改造工程造价快速测算提供了一种新方法.
Cost Prediction of Existing Building Renovation Based on PCA-BP Neural Network
Due to construction process information has been missed and damaged for existing build-ings,as well as the lack of basic cost estimation standard data such as cost quota data.It is difficult to quickly and accurately calculate the cost of renovation project during the decision-making stage.To adress this issure,a method of predicting the cost of existing buildings renovation based on principal component analysis(PCA)and BP neural network is proposed.According to the case and literature ayalysis,10 influencing factors that affect the cost of existing building renovation were identified and extracted.By utilizing principal component analysis,the extracted 10 factors were reduced dimension-ality and 3 new comprehensive variables were identified.BP neural network was used to train,validate and test 280 existing building renovation costs in groups.The PCA-BP neural network model is based on reduction and elimination of correlation dates among indicators.The results show that the PCA-BP neural network model improves higher processing and prediction efficiency based on dimen-sional reduction and elimation of correlation data between indicators.It achieves good results in root square error and correlation.The comprehensive error between the measured data and the predicted data was 2.66%.This paper provides a new method for rapidly estimation of renovation of existing construction projects.

existing buildingPCABPengineering costpredict

赵伟佳、罗德才、陈方、陈倩

展开 >

湖北文旅园区建设发展集团有限公司,湖北 武汉 430071

中国船舶集团有限公司第七二二研究所,湖北 武汉 430200

北京市市政工程设计研究总院有限公司,北京 100082

既有建筑 PCA BP 工程造价 预测

2024

土木工程与管理学报
华中科技大学

土木工程与管理学报

CSTPCDCHSSCD
影响因子:0.837
ISSN:2095-0985
年,卷(期):2024.41(2)
  • 13