【作者】 刘斌；

【导师】 胡庆国；

【作者基本信息】 长沙理工大学 ， 管理科学与工程， 2021， 硕士

【摘要】 随着我国高速公路的快速发展,一起而来的还有投资失控现象。高速公路项目前期的投资估算可以为项目可行性研究和设计方案比选提供参考,也可以作为高速公路后期各阶段造价控制的重要依据。另外大数据时代的到来,工程样本数据的获取也将更为方便。因此本研究针对高速公路的投资失控现象,利用计算机技术和数学估算模型,以已竣工的工程数据作为样本库,构建既快速又准确的投资估算模型具有重要意义。论文主要工作如下:首先,梳理了高速公路项目工程造价的概念以及组成,并阐述平均影响值算法(MIV)、粒子群算法(PSO)以及支持向量机算法(SVM)的选取理由及其相关理论知识。其次,通过分析高速公路造价的特征影响因素并查阅以往相关研究资料,初步选取高速公路工程造价估算敏感性特征指标,并且划分定性指标与定量指标,以及定性指标的量化处理。然后,构建MIV-PSO-SVM投资估算模型,通过不同的核函数训练模型找到最适合高速公路造价估算的核函数。按照逐步剔除影响最小指标的原则选择最优敏感性特征指标体系。最后详细介绍了本研究所提出模型的MATLAB实现方法并通过已竣工的工程数据进行仿真分析,仿真结果表明,相较于BP神经网络、SVM以及传统造价估算方法,本研究构建的估算模型预测精度更好、速度更快且效果更稳定,可以满足高速公路投资估算的精度要求。本研究构建的MIV-PSO-SVM高速公路投资估算模型,基于MATLAB软件算法开发和数据分析的交互式环境,具有运行速度快、预测精度高且模型简单易懂等特点。可以为工程项目前期的决策以及后期各阶段的造价控制提供依据,也可以为同类工程的投资估算提供参考,具有一定的实践价值和理论价值。更多还原

【Abstract】 With the rapid development of highways in our country,there is also the phenomenon of out-of-control investment.The investment estimate in the early stage of the expressway project can provide a reference for the feasibility study of the project and the comparison and selection of design schemes,and can also be used as an important basis for the cost control of the expressway in the later stages.In addition,with the advent of the big data era,the acquisition of engineering sample data will be more convenient.Therefore,this research aims at the phenomenon of expressway investment out of control,using computer technology and mathematical estimation models,and using completed engineering data as a sample database to construct a fast and accurate investment estimation model.The main work of the paper is as follows:First of all,it sorts out the concept and composition of highway project engineering cost,and elaborates the reasons for the selection of average influence value algorithm(MIV),particle swarm algorithm(PSO)and support vector machine algorithm(SVM)and related theoretical knowledge.Secondly,by analyzing the characteristic influencing factors of expressway cost and consulting related research data in the past,we initially select the sensitive characteristic index of expressway project cost estimation,and divide the qualitative index and quantitative index,as well as the quantitative treatment of qualitative index.Then,construct the MIV-PSO-SVM investment estimation model,and find the most suitable kernel function for highway cost estimation through different kernel function training models.The optimal sensitivity characteristic index system is selected according to the principle of gradually eliminating the least influential indexes.Finally,the MATLAB implementation method of the model proposed by this research is introduced in detail and the simulation analysis is carried out through the completed engineering data.The simulation results show that compared with the BP neural network,SVM and traditional cost estimation methods,the estimation model constructed in this research predicts The accuracy is better,the speed is faster and the effect is more stable,which can meet the accuracy requirements of highway investment estimation.The MIV-PSO-SVM neural network highway investment estimation model constructed in this research is based on the interactive environment of MATLAB software algorithm development and data analysis.It has the characteristics of fast running speed,high prediction accuracy and simple and easy to understand model.It can provide a basis for the decision-making in the early stage of the project and the cost control of the later stages,and it can also provide a reference for the investment estimation of similar projects.It has certain practical and theoretical value.更多还原