| The development of the construction industry has improved the quality of people’s life,but it has also brought a problem that cannot be ignored--construction safety accidents.The construction process is characterized by high personnel mobility,various and complicated operation types and manual operation,which leads to frequent construction safety accidents.Safety assessment is considered to be an effective method to control safety accidents.Therefore,in this paper,safety evaluation methods are studied for the construction environment in order to reduce the casualties in construction.Based on Heinrich’s accident causation theory,a construction safety evaluation index system is constructed by finding locations of high-frequency safety accidents.Starting from the sudden characteristics of safety accidents,the catastrophe theory is introduced to generate a construction safety mutation model.On the basis of traditional catastrophe evaluation,by constructing the reverse mapping relationship between the underlying index and the membership degree of the target layer,the problem that the membership of the system total catastrophe does not have evaluation significance is solved,and the improved catastrophe evaluation method is obtained.The evaluation result is consistent with the actual safety state.In order to conduct safety evaluation conveniently and quickly,considering the advantages of self-learning of the BP neural network,it is combined with the improved catastrophe evaluation method.By taking the raw data of the underlying indicators as input and the comprehensive evaluation value obtained by the improved catastrophe evaluation method as the output,a BP-building construction catastrophe evaluation model with rapid evaluation of the safety status is constructed.Experiments found that the evaluation results have a problem of excessive sample errors,and the minimum mean square error obtained when Epoch=17 cannot fully meet the highprecision requirements of building construction evaluation,so genetic algorithm is introduced to optimize it,and a BP neural network building construction safety evaluation model optimized by genetic algorithm is constructed.Experiments show that this model can accurately evaluate the building construction process. |
