Building Safety Accident Prediction Based On Fuzzy Cognitive Map

At present,the low level of safety in China's construction industry runs counter to the requirements of building high-quality engineering goals.The situation of construction safety production is facing challenges.At the same time,it also restricts the rapid development of the construction industry.Therefore,it is urgent to improve the safety production situation in the domestic construction industry.Level is of great significance.The construction industry projects are large in scale,long in construction time,relatively complex in construction environment and uneven in overall quality of operators.Compared with other industries,there are more uncertainties and high risks in the implementation of construction projects.The construction site needs to pay attention to and prevent the occurrence of construction safety accidents,and it is necessary to find out what causes the construction safety accidents.Many different factors have different influences.The factors that cause construction safety accidents are often not single.Therefore,a good algorithm model is needed when analyzing and predicting construction safety accidents.This article first obtains 298 construction safety accident reports through text mining technology,extracts the keywords in the report,obtains the number and frequency of different keywords,and classifies the extracted keywords according to the construction safety inspection standards.Then use genetic algorithms to train the obtained keyword data,and obtain the causality and influence value between different factors through training.These factors are human factors,physical factors,environmental factors,management factors,and casualties.The causal relationship between factors and the size of the influence value construct a weight matrix.In the weight matrix constructed,there is a causal relationship between the factors of things and casualties,and it has the greatest influence value.This may be because the keywords for things factors in the construction safety accident report are mentioned more frequently than those for other factors,It suggests that the things factors have a greater impact on the casualties of people.Finally,we build a fuzzy cognitive map(Fuzzy Cognitive Maps,FCM)model to predict construction safety accidents.The fuzzy cognitive graph model can analyze the interrelated scenario of many different factors,and can show the causality and influence value between various factors,which is also different from other machine learning algorithm models.After 31 iterations of the FCM model proposed in this paper,the system tends to be stable.At this time,the state values of all factors remain unchanged.From the analysis of the experimental results,we can see that the management factor is the leading factor in the construction safety accident.By modeling and analyzing examples of construction safety accidents,management factors have become the most important factors for construction units,and reminded construction unit managers to strengthen management,make arrangements for construction workers according to the actual situation,and regularly train and improve construction workers.The safety consciousness would kill the signs of accidents in the cradle and reduce the occurrence of construction safety accidents.