Case-Based-Reasoning On Safety Accidents In Super-high-rise Buildings Based On Random Forests

In today's society,the development of super high-rise buildings is very rapid,such as bamboo shoots after a spring rain,but with it comes the frequent safety accidents of super high-rise buildings.Due to the particularity of super high-rise building itself,once the accident happens,The severity and influence degree are huge.However,similar safety accidents of super tall buildings are always repeated,and how to break the "magic spell" of frequent accidents has become a topic that cannot be ignored in the whole industry.High-rise buildings,the author of this paper the special architectural forms,the application of "heavy experience,weak theory" of case-based reasoning model,build up an index system of high-rise building safety accidents,develop a number of 140 super-tall building safety accident case,design index of similarity and rule for calculating the case similarity of the query is most similar to the case;With the random forest model as the basic learner,the two-layer random forest model was constructed.The indexes of the query cases were input and the auxiliary decision-making of the case was output.The reuse model was constructed,and the model was verified by the examples.The main research results of this paper are as follows:(1)Through literature mining and analysis of the influencing factors of ultra-high rise building safety accidents,and from the accident characteristics,project characteristics and organizational characteristics of three dimensions to establish the ultra-high rise building safety accident characteristics index system;Collect and sort out the safety accident cases of super high-rise buildings,and build the relevant case database,study each accident rectification report,and sort out the accident rectification indicators.(2)A case-based reasoning model for safety accidents of super high-rise buildings is constructed.Ant colony clustering method is introduced in the case inquiry stage to cluster cases in the case database,forming case case database and improving the accuracy and efficiency of case inquiry.The similarity index system of safety accident of super high-rise building and the calculation method of case similarity are constructed.In the case reuse stage of case-based reasoning model,the idea of ensemble learning algorithm is adopted to construct a two-layer training model for auxiliary decision making of safety accident correction index of super high-rise building.The random forest model is selected as the integrated learning model of the two-level training model.(3)With Python programming technique,the CBR model of safety accident of super high-rise building built in this paper based on double-layer random forest was implemented,and the visual case input in the case inquiry stage was realized with interface.