Research On Safety Early Warning Of Construction Engineering Based On Convolution Neural Network

The construction industry plays an important role in the process of economic development in our country,but the construction safety accidents that happen frequently also cause widespread concern of the society and academia.The high number of annual safety accidents and casualties reflects that there are some problems in the current construction safety management,such as the slow detection of safety accidents and the untimely warning of safety accidents.China’s construction industry urgently needs new safety management methods and technologies to reverse the current inefficient safety management mode of the construction industry.At present,the use of new technology for intelligent construction safety management results less.In this paper,convolution neural network is used to build a video image recognition and classification model to realize the early warning of dangerous area invasion and unsafe behavior in the construction site,reduce the corresponding safety accidents,and then carry out empirical analysis.The premise of realizing intelligent early warning of construction site is to obtain the influencing factors of construction site insecurity.In this paper,firstly,the literature review method is used to get the influencing factor system of the unsafe construction site,and then the empirical analysis method is used to calculate the coupling of several kinds of unsafe factors with the greatest risk of the construction site in China,which provides the focus for the following safety early warning.Based on the analysis of relevant documents and standards on unsafe behaviors of construction personnel,this paper obtains 13 types of unsafe behaviors of construction personnel;combined with the existing research and analysis,three types of dangerous areas are found,namely,the excavation construction dangerous area,the mechanical movement dangerous area and the electric power leakage dangerous area,and delimits the dangerous area delimitation rules.In this paper,neural convolution network is introduced to realize the early warning of invasion behavior in the dangerous area of construction and the early warning of unsafe behavior of personnel in the dangerous area of construction site.In this paper,the early-warning principle of invasion behavior in dangerous area and the early-warning principle of unsafe behavior of personnel in dangerous area of construction site are put forward.Finally,the actual construction site cases are used to verify the early-warning process of the two models,to show the early-warning effect and early-warning accuracy.In this paper,convolution neural network is used to forewarn two kinds of construction site safety accidents,which can reduce the probability of this kind of safety accidents to a certain extent,ensure the safety of the field staff,and provide new ideas for further realization of intelligent construction site,which has important theoretical significance and application value.