Study On Cost Estimation Of Urban Rail Transit Project Based On BP Neural Network

Cost estimation of urban rail transit engineering is an important basis for investment decision of urban rail transit engineering construction.Accurate and fast cost estimation directly determines the correctness of investment decision.At the same time,it is also the decisive factor for the success of cost control in the later stage of project engineering.However,at present,in the field of urban rail transit engineering construction,engineering cost estimation mainly relies on various computer cost software.This method has high requirements for the design of decision-making stage,relies on the support of more resources,and requires a relatively long time.At present,more and more second-and third-tier cities that lack rich construction experience join the ranks of new urban rail transit projects.If the owners of these cities are unfamiliar with urban rail transit construction,or can provide limited resources,or the design ability of the design institute is insufficient,it is easy to use engineering cost software to predict engineering cost,which leads to the problem of long estimation period and low accuracy of engineering cost estimation.The purpose of this study is to achieve an accurate and fast cost estimation of urban rail transit engineering in the early stage of project construction when the acquisition of engineering characteristic data indicators is limited.By constructing the index system of project cost estimation,BP neural network algorithm is applied to the case study of urban rail transit project cost prediction.The research results show that the BP neural network model can achieve more accurate and fast cost estimation of urban rail transit project with relatively less resources and time.The research results can provide accurate and rapid reference for the investment decision of urban rail transit engineering construction projects(especially for some second-and third-tier cities that lack rich construction experience),and have a certain practical significance for urban rail transit construction investment projects.Firstly,this paper comprehensively analyzes the influencing factors of project cost based on software quota analysis.Considering the reality of the early stage of the project engineering degree of available features,combined with the expert interview law and the principle of index system building,optimize the based on the norm analysis of influencing factors,will not be important,repetition and actual pre-project are difficult to obtain indicators shall be simplified,based on the BP neural network is constructed the index system of urban rail transit construction cost,The index system is simple,practical and effective.Secondly,according to the operation mechanism of BP neural network and the experience basis of the framework of neural network,the cost model of urban rail transit project based on BP neural network is established.The main engineering characteristic factors of the index system constructed in this paper are the main influencing factors of urban rail transit as the input variables and the cost of urban rail transit as the output variables.The function estimation relationship between the main engineering characteristics of urban rail transit and the cost of urban rail transit is established.The BP neural network algorithm has the adaptive learning property.It can process the nonlinear data quickly and accurately,and realize the accurate and rapid estimation of the cost of urban rail transit project.Finally,using the above method based on BP neural network model of urban rail transit construction cost estimate of actual engineering case for project cost estimate an empirical case analysis,draw a percentage of the data as training samples for BP neural network training,the remaining samples for BP neural network prediction,prediction,Then the difference between the output value and the actual value is tested to verify the effectiveness of the BP neural network model for estimating the cost of urban rail transit project.The results show that the BP neural network algorithm can realize the accurate and fast construction cost of urban rail transit.