PREDICTION OF BUCKLING LOAD OF COLUMNS USING ARTIFICIAL NEURAL NETWORKS

A number of investigators have proposed semiempirical formulas for the critical buckling load of slender columns. The departure from the ass umptions of the elastic-plastic theory makes the task of incorporating all the features of real-life columns into a single formula very diff icult. As a result, semiempirical formulas, adopted for design specifi cations often follow a lower bound to experimental observations to inc lude a variety of column types. Therefore, a significant portion of th e actual column strength remains unutilized, when such a lower bound i s adopted in the design of axially compressed members. This technical note reports development of a tool for the prediction of buckling load of columns, which requires minimum assumptions using neural computing techniques. This concept can be extended to include a variety of colu mn types in a single model for the buckling load of columns. This conc ept can also be further extended for reliability analysis as the netwo rk can also predict the standard deviation in the column strength.