Research On Risk Theme Mining And Correlation Visualization Method Of Subway Deep Foundation Pit Construction

Since the 21 st century,China’s urban rail transit infrastructure construction efforts continue to increase,subway station deep foundation pit construction safety accidents occur frequently,indicating that there are defects in site construction risk management.At present,most scholars use structured data such as monitoring data,expert rating or analogue simulation to carry out construction risk research,but rarely use unstructured text data to study risk,and construction risk records accumulated in engineering practice have not been mined and utilized.Therefore,this article innovatively puts forward the subway deep foundation pit construction risks theme mining and correlation visualization method,base data on risk record texts collected from engineering project practices,dig up the implicit risk themes and correlation,draw visual network diagram of construction risk correlation,to improve the efficiency of identifying and acquiring risk associated information,it is that scientifically and accurately guide the construction party to take preventive and control measures to manage key risks and associated risks,and improve the field construction risk management.Its main research contents include:First,put forward the research questions,discuss the relevant theories and methods,and sort out the research ideas.From the background of subway construction and the analysis of relevant research status at home and abroad,there are shortcomings in the risk management of subway deep foundation pit construction.In particular,most of the researches use structured data to carry out construction risk assessment and analysis,and a large number of accumulated text data of construction risk cases have not been mined and utilized.From the research question,discuss the risk management theory of subway deep foundation pit construction,text mining and visualization related theories and methods,and the research ideas and contents of this paper are sorted out and discussed,lay a theoretical and methodological foundation for the research on theme mining and correlation visualization methods of subway deep foundation pit construction risk.Secondly,discuss and test the text data preprocessing of subway deep foundation pit construction risk cases text.Jieba toolkit in Python is used for Chinese word segmentation of construction risk case text,adding user-defined dictionary and deactivated dictionary to optimize and improve word segmentation effect,so as to realize text data corpus cleaning and filtering.Vector space model(VSM)is used to realize text vector transformation,TF-IDF algorithm is introduced to weight calculation,and high-weight construction risk feature words are retained to achieve text data reduction and denoising.Finally,a sample experiment is used to show the text data preprocessing process,which provides data basis for the construction risk theme mining and correlation visualization research.Thirdly,the subway deep foundation pit construction risk theme is carried out with text mining model.A risk theme text mining model based on LDA theme model and K-means text clustering is constructed to mine construction risk theme from a large number of risk case text data,to realize key risk mining analysis in big data.The semi-automatic method was used to allocate risk record labels,calculate the accuracy rate,recall rate and F value of the evaluation index of text mining results,analyze the accuracy rate of construction risk theme mining over 90%,verify the applicability and feasibility of the model,and verify the scientific nature of the basic data and the reliability of mining results.By further comprehensive analysis of the text mining results,30 construction risk themes are obtained,and the key risk points that should be focused on investigated are summarized,so as to guide the investigation and control of construction risks in subway deep foundation pit construction and ensure site construction safety.Finally,the risk correlation visualization method of subway deep foundation pit construction is established.First use of Apriori algorithm for mining correlation rules in the construction risk,risk themes was used to optimize and improve risk associated visualization network diagram,efficiently and intuitively analysis the risk correlation relationship,take measures to prevent and control risk correlation rules former term,effectively reduce the incidence of the correlation rule risk after item,improve site construction risk management.Then social network analysis(SNA)is used to explore and analyze the correlation between risk and time and space,utilize the visualized theme optimization to improve risk associated network diagram,improve the efficiency of the identification for risk correlation and accuracy,and so as to take preventive and control measures to manage correlation risks,effectively improve the risk management of construction.This paper carries out research on risk theme mining and correlation visualization methods of subway deep foundation pit construction,and realizes the method integrated application and improvement innovation,with the LDA theme model,K-means text clustering,Apriori algorithm,social network analysis(SNA)and knowledge visualization methods.From a large number of construction risk case text data dig up the implicit in the themes and the risk correlation,draw visual network diagrams of risk correlation,improve the efficiency and accuracy of identifying and acquiring construction risk related information,not only to verify the applicability and feasibility on risk theme mining and correlation visualization method of subway deep foundation pit construction,but also it can guide the site construction to strengthen the prevention and control of key risks and associated risks,and supplement and improve the site safety management by using the results of big data analysis,which has important guiding significance for the risk management practice of subway deep foundation pit construction.