An Early Warning System For Construction Safety Based On Bayesian Network

Nowadays, construction industry has become one of the most important domestic primary breadwinner, because of the dispute fact that the influence of construction industry is strong and covers a wide scope. With the development and progress of science and technology, methods of safety management in construction industry is gradually emerging from traditional ones in order to have a better development in industry field. More and more scientific safety management methods have been introduced into the construction field. At present, there are many researches on construction safety management and early warning system both at home and abroad. These researches have achieved certain results, which have positive effects on the current construction safety warning system. However, at the present stage, these studies are still lack of the high precision and strong timeliness warning model. Therefore, it is essential to build a practical, timeliness and accurate warning model to meet the target that aims to transform the post-processing to early-warning.Firstly, this dissertation carries on a thorough analysis of current construction safety situation and points out that because of the current high death rate, it is necessary to make a deep research on early-warning system. Secondly, combining the accident-causing theory and the characteristics of construction, this dissertation makes an investigation on factors of construction safety. Besides, an early-warning index system is established through the data analysis and screening. Thirdly, with the help of Software Ge NIe2.0 which is designed to build the Bayesian Network,this dissertation creates a Bayesian Network Model based on the information gathered from the construction site and experts’ experiences, which can be applied into the construction safety early-warning system. Moreover, the established model can find out key influencing factors through backward inference and sensitive analysis. Last but not the least, given the function and operation structure, this dissertation makes a detail depict of the establishment of the early-warning system based on the Bayesian Network. Meanwhile, to prove the feasibility of the established system, a living example is introduced to illustrate the operational process and the result highly lives up to the expectation.