Research On Dynamic Risk Assessment Model For Fire Incidents In Commercial Premises

More and more people are moving into urban life,and the people’s desire is to live a good life.This means a significant increase in commercial premises.The structure of commercial places is complex,and some commercial places have large volumes,dense personnel,high fire load density,and difficult evacuation,which pose a threat to people’s safety.Therefore,it is of great significance to conduct fire risk assessment on commercial premises and provide data support for later key maintenance.Traditional fire risk assessment methods are no longer sufficient to meet the current development needs in the context of increasing complexity and number of commercial sites.The dynamic fire risk research model of commercial premises developed in this thesis is helpful for guiding the reduction of fire risk in commercial places.In order to study fire risk assessment,it is necessary to establish an index system library covering perfect indicators.Taking a large number of commercial places in the city as the research object,this thesis analyzes the influencing factors of fire occurrence and the fire risk characteristics of different types of places by studying fire cases,so as to determine the index system library applicable to different types of places,which lays an important foundation for subsequent fire risk assessment.In this study,based on the index system database established in the previous period,through the obtained fire statistics and expert suggestions,the indicators were screened from the index system library to match the specific commercial places,so as to sort out the model training sample data.The BERT model is used to train the training sample data to automatically generate a matching index system according to the different characteristics of the site,that is,the adaptive index system.The BERT model has powerful natural language processing capabilities,and can learn richer text representations through large-scale unsupervised pre-training,so as to achieve excellent results in various natural language processing tasks.In this study,a site adaptive index system model was constructed using the Bert model.The prior probability of the index is calculated by using two algorithms,maximum likelihood estimation method and fuzzy theory,and a Bayesian model is established to predict the probability of fire.The conditional probability is calculated by the weight allocation method,and the Bayesian algorithm model is obtained to calculate the size of the fire value in commercial places.The final commercial site fire dynamic risk research model is designed and developed by the Py Charm tool.Finally,the index system can be automatically generated according to the user’s demand input conditions,and the dynamic curve of fire risk can be calculated through Bayesian algorithm.The results of the developed commercial site fire dynamic risk research model are compared and analyzed with the model made by Genie Software.The application results show that this model has good results,which can provide certain help for the analysis and judgment of urban fire protection work.