Research On Cost Forecast Of Tunnel Engineering Based On Historical Data

The rapid increase in demand for infrastructure construction in the process of urbanization has led to a serious shortage of urban land resources,and the contradiction between space supply and demand has become increasingly prominent.Relevant government departments have gradually paid attention to urban underground space.The number of tunnel projects has increased year by year,but it has also been accompanied by cost overruns.One of the reasons for this serious problem is the underestimation of the tunnel project cost forecast.Due to the lack of detailed data at the design stage,the total cost of the project could not be accurately estimated.At the same time,in the feasibility stage,the cost estimation of the tunnel project can be used as a basis in the investment decision of the scheme.Therefore,this article summarizes the main influencing factors of tunnel project cost underestimation through the collection of historical tunnel project data,and discovers the correlation between tunnel cost and tunnel cost influencing factors through exploratory analysis of data,and uses four algorithm designs to construct tunnel projects Cost prediction model to achieve accurate prediction of tunnel project cost.According to the literature search and tunnel cost underestimation influencing factors summary,forty influencing factors were summarized and ranked according to their frequency in the literature.Finally,the top four influencing factors were selected as tunnel length,tunnel diameter,and environment.Factors and geological conditions.Through data collection and standardization,exploratory analysis of the collected data,including descriptive statistics and data fitting,etc.,found the data differences between hard rock and soft rock,and summarized the tunnel cost and tunnel diameter as well as The different relational equations of tunnel length under hard rock conditions and soft rock conditions are fitted and evaluated according to the evaluation index R~2.Using logistic regression,decision tree,neural network and GBDT four algorithms,the tunnel length and tunnel diameter are used as input feature factors,through the design of the algorithm,the hard rock and soft rock tunnel cost prediction models are established respectively,and the determination coefficient R~2 is used to pair The results are analyzed and evaluated.These four algorithms can better predict the cost of tunnel engineering.Among them,the neural network algorithm has the best prediction effect on hard rock tunnel projects,and the Logistic has good prediction effects on soft rock tunnel projects,and hard rock tunnel projects are generally better than soft rock tunnel projects.