Research On Office Building Cost Estimation Based On Machine Learnin

When making investment decision in the early stage of construction,the estimated value of project cost is the key reference data,which is the basis for the preparation of project proposal and feasibility study report,and runs through all stages of subsequent project implementation.Accurately and efficiently estimating the engineering project cost has always been the concern and research focus of practitioners,and the accuracy of the estimation will directly affect the investment decision of the project,which is crucial for all parties involved.With the flourishing development of machine learning,various intelligent algorithms have been applied in the field of engineering cost estimation in order to estimate cost quickly and accurately.This study selects office buildings as the research object and chooses suitable machine learning methods,aiming to build an engineering cost estimation model with an error of less than 10% for cost management at the early stage of engineering projects,and the main research contents are as follows:First of all,on the basis of analyzing the current situation of domestic and international research and summarizing the cost estimation methods,the theories related to office construction,engineering cost and machine learning are explained.In view of the incomplete information at the early stage of the project,the indexes reflecting the characteristics of engineering "quantity" and "price" are selected and adjusted by filtering through the literature.The "quantity" of the project is reflected in the scale and features of the project.The indexes of building area,number of floors and floor height can directly reflect the scale of the building,while the indexes of structure type and foundation form can reflect the features of the project.The new construction area of office buildings,regional GDP and consumer price index can reflect the influence of market economy on the "price" of the project.Based on the combination of construction engineering indexes and market economy indexes,the initial indicator system is constructed.Then principal component analysis is applied to reduce the dimensionality of indicator system,eliminating the duplicated information among indicators,and generating the input vector of the estimation model.Secondly,the BP neural network model and SVR model were constructed by reasonably selecting training parameters and kernel functions.A total of 85 office buildings in 19 regions are processed and input into the model for training,and the performance of the model is evaluated and compared.The results show that the estimation error of the BP neural network model is about 12%,which does not meet the expectation,and the estimation error of the SVR model is about 8%,which is controlled within 10% and meets the accuracy requirement.After investigation,it is known that the BP neural network uses gradient descent method to update the weights and thresholds,which is easy to fall into the local optimum situation and trigger the network instability.In order to improve the accuracy of the existing BP neural network model,the genetic algorithm is used to optimize the BP neural network,so that the GA-BP neural network model is constructed which has a faster convergence speed and stronger generalization ability.The estimation accuracy of the GA-BP neural network model is the highest,with an error of about 6%,which effectively improves the accuracy of office building cost estimation.Making full use of the data information and machine learning algorithms can improve the accuracy and efficiency of cost estimation and realize the intelligence of office building cost management.With the assistance of machine learning,it can enable the managers to plan and utilize the project funds reasonably and safeguard the interests of all parties involved in the project while contributing to the dynamic management of project costs by converting post-facto control of construction projects into pre-facto control.