Construction Production Safety Accident Cause Correlation Analysis And Trend Froecasting Research

As one of pillar industries of national economy in our country,the construction is employed the largest number of industries,but due to its complex production environment,production safety accidents occur frequently,which make the state property and people’s life and property serious loss.But now the research of construction safety accidents more is on systematic analysis of the safety management level and some indicators of statistical analysis,the lack of internal causes of accident analysis.This paper introduces the data mining theory knowledge to system modeling for construction safety accident,combined with 2004 to 2014,the national construction accident data and the specific situation of the Beijing municipal construction accident data for trend prediction of construction safety accidents and accident factors and causal relationship between the empirical analysis.The specific work carried out the following three aspects:1.Construction production safety accident trend forecasting research.This part mainly to the national construction accident quarterly data ARIMA model and Holt-Winter exponential smoothing model is set up to forecast,and through to evaluate two models,and the comparison of error rate,will pick the best model to forecasting on the future trend of architectural accident.2.Construction accident factor correlation analysis.By using the association rules in data mining to Beijing construction accident data to construct the relation network,mainly made by accident of casualties against doing a correlation analysis,by optimizing the screening of mining frequent itemsets of support and confidence,to judge the relationship between the cause of the accident,to draw in the Beijing area construction characteristics of the root cause of the accident.3.Construction accident factors between causal analysis.By using the complex network building construction accident factors cause net,based on the network topology characteristics,the causality between the mining accident factors.This part is mainly to causality between before and after the cause of the accident factors with complex networks of knowledge,so as to let us to cause construction accident factors have intuitive,holistic understanding.And ideas will be discussed to establish the national construction safety accidents in the past ten years data and construction safety accident in the Beijing area on the basis of objective,in fact,the development trend of construction accidents and accidents caused by factors,internal connection between the respectively using the time series prediction,and complex network related knowledge of data mining association rules has carried on the empirical analysis,provide preventive measures and guidelines for safety accident happened.