Assessment of the ultimate capacity of the cellular steel beams subjected to the lateral-torsional buckling containing web distortion (lateral-distortional buckling) was rarely investigated. So far, there is no accurate formula to predict the lateral-distortional buckling (LDB) capacity of such members. Here, extensive simulations (660 specimens) based on the nonlinear finite element analysis (FEA) were carried out to simulate the cellular steel beams in distortional-buckling mode. The influence of eight parameters that may affect the LDB capacity was deeply investigated. Then, gene expression programming (GEP) and nonlinear multiple regression (NMR) algorithms progressed as predictive models to obtain formulae for measuring the failure moment capacity of the cellular steel beams. The used data for the GEP and NMR models development, obtained from the FEM simulations that were done initially. It was attempted to derive exact enough formulae, which considers all the variables that may affect the LDB capacity of cellular steel beams. The validation of the GEP and NMR models have been further verified by several statistical criteria. Finally, a comparison was performed between the proposed formulae and the predictions obtained from AS4100, EC3, and AISC codes. It was shown that the proposed formulae are enough exact and applicable to use by practicing engineers.

很少研究对包含腹板变形（横向变形屈曲）的横向扭转屈曲的蜂窝状钢梁的极限承载力的评估。到目前为止，尚没有准确的公式来预测此类构件的横向变形屈曲（LDB）能力。在此，基于非线性有限元分析（FEA）进行了广泛的模拟（660个样本），以变形屈曲模式模拟了蜂窝状钢梁。深入研究了可能影响LDB容量的八个参数的影响。然后，基因表达编程（GEP）和非线性多元回归（NMR）算法作为预测模型进行了改进，从而获得了用于测量蜂窝钢梁破坏能力的公式。用于GEP和NMR模型开发的数据，从最初完成的FEM模拟获得。已尝试得出足够精确的公式，其中考虑了可能影响蜂窝钢梁LDB承载力的所有变量。GEP和NMR模型的验证已通过几种统计标准进一步验证。最后，在提议的公式与从AS4100，EC3和AISC代码获得的预测之间进行了比较。结果表明，所提出的公式足够精确，适用于实践工程师使用。在提议的公式与从AS4100，EC3和AISC代码获得的预测之间进行了比较。结果表明，所提出的公式足够精确，适用于实践工程师使用。在提议的公式与从AS4100，EC3和AISC代码获得的预测之间进行了比较。结果表明，所提出的公式足够精确，适用于实践工程师使用。