The web tapered I-section steel (WTIS) columns have been widely used in civil and industrial steel structures. However, the existing theoretical and empirical equations demonstrate a significant discrepancy in estimating the critical axial load of the WTIS columns. This study aims to develop effective artificial neural networks (ANNs) for predicting the critical buckling load of the WTIS columns. A database of 269 finite element models of WTIS columns was generated, after verifying with experimental results, to develop the ANN model. The results of the proposed ANN model were also compared with those of existing formulas, highlighting that the ANN model in this study predicts the critical buckling load of the WTIS columns more accurately than the existing formulas. Moreover, the influences of input parameters on the critical buckling load of the WTIS columns were thoroughly investigated. An ANN-based formula, which considers input variables, was thereafter proposed to estimate the critical buckling load of the WTIS columns. Additionally, a graphical user interface tool has been developed for simplifying the design practice of the WTIS columns.

腹板锥形工字钢（WTIS）柱已广泛用于民用和工业钢结构。然而，现有的理论和经验方程表明，在估算 WTIS 柱的临界轴向载荷时存在显着差异。本研究旨在开发有效的人工神经网络 (ANN) 来预测 WTIS 柱的临界屈曲载荷。建立了 269 个 WTIS 柱有限元模型的数据库，经过实验验证，建立了 ANN 模型。所提出的 ANN 模型的结果也与现有公式的结果进行了比较，突出表明本研究中的 ANN 模型比现有公式更准确地预测了 WTIS 柱的临界屈曲载荷。而且，深入研究了输入参数对 WTIS 柱临界屈曲载荷的影响。此后提出了一个考虑输入变量的基于 ANN 的公式来估计 WTIS 柱的临界屈曲载荷。此外，还开发了一个图形用户界面工具来简化 WTIS 柱的设计实践。