The factors,which influence the stability of thin-wall perforated steel components are complex. Take the columns as an example,in the existing research,much attention has been paid to the continuous section,but little paper on noncontinuous one. According to the characteristics of the non continuous cross section,based on performance testing and finite element analysis of columns,with the method of BP neural network intelligent stability prediction,the stability prediction model of the columns is established. The predict results are compared with the test data,and the error is less than 2%,which is more accurate than the finite element analysis results. It is very promising that the intelligent prediction method has applied in the structure engineering.