Research On Case Analysis Of Tower Crane Accidents Based On Data Mining

As a common lifting and transportation equipment in construction,tower crane is widely used and accidents occur frequently.It is difficult for safety management personnels to manage the tower crane because of its special structure,complex accident formation mechanism and uncertain accidents.Therefore,based on the tower crane accident investigation reports,this thesis proposed a unified analysis framework,structured extraction of accident information and combined data mining technology to explore the accident rules deeply.In addition,the model of tower crane accident prediction and causation was established,which provides the basises for the safety management decisions of tower crane.Firstly,218 tower crane accident investigation reports were collected,and a systematic tower crane accident causation analysis model was established based on24 Model.A unified analytical framework was proposed and tower crane accident cases were extracted with WBA method.Based on this,an improved Apriori algorithm suitable for multi-dimensional tower crane accidents data was proposed to improve the efficiency of accident mining.And then,the correlation between the basic information of tower crane accidents,the types of accidents and the causes of accidents was quantitatively excavated,and the targeted safety management suggestions were put forward.Secondly,based on support vector machine,a tower crane accident type prediction model was established to predict the most likely type of tower crane accident by using the information easily obtained on site and reduce the volume of causal analysis.Then,according to the prediction results,the corresponding Bayesian network causative analysis model was established to determine the ordering of each causative node and guide the managers to formulate reasonable security management strategies.Finally,the model of tower crane accident type prediction and causation was demonstrated and verified through 5 newly collected tower crane accident cases,and the model has a good effect.The unified report analysis framework in this paper can standardize the extraction of tower crane accidents information,the improved Apriori algorithm can dig the tower crane rules deeply,the tower crane accident type prediction and causation model can guide the safety management strategy.Furthermore,this thesis also provides a reference for data mining method in accidents analysis and application.