Analysis and prediction of substation construction safety accidents based on CART model
In the current substation construction process,the data envelopment analysis process is mainly used to predict the safety accidents,and the uncertainty in the characterization information is ignored,which leads to the low receiver operating characteristic curve area(AUC)value in the selection of the prediction results.In order to solve this problem,a new analysis and prediction method of substation construction safety accidents is designed by using classification and regression tree(CART)model.Firstly,the data of substation construction site are collected by combining fixed and mobile acquisition technologies,and filtered by principal component analysis algorithm.Then,the occurrence process of substation construction safety accidents is deeply analyzed,and the precursor characteristics of construction safety accidents are extracted through the separability criterion based on probability distribution.Finally,the CART model is used to build the root node of construction safety accidents,and then the support vector machine(SVM)regression algorithm is used to build the leaf node,forming the optimal decision tree that can be used for construction safety accident prediction.By iteratively training multiple CART models in series,the gradient can be im-proved,and accurate accident prediction results can be obtained by applying this model.The experimental results show that the prediction method is more sensitive and can predict more safety accidents,and the AUC value of the prediction method is as high as 0.91,which has higher prediction accuracy.
classification and regression treesubstation constructionsafety accidentspredictionfeature classificationsupport vector machine
万方数据
维普
