Shear performance prediction of hollow slab beams in concrete buildings based on improved GEP

Abstract The shear performance of concrete buildings is an important component of building safety. Predicting and analyzing shear performance can effectively reduce safety accidents. However, existing methods have problems such as large prediction errors. Therefore, a method for predicting the shear performance of hollow slab beams based on improved gene expression programming is proposed. This method adopts a pyramid evolution strategy for layered search, with different evolutionary probabilities set at different layers. The semi-elite roulette wheel is used for initial population selection, the neighborhood optimization is taken to improve mutation strategy, and a promotion mechanism is set. Finally, the prestress, shear span ratio, and rust factors are used to improve the correction coefficients of the shear performance calculation equation. The experiment showed that the convergence speed of improved gene expression programming was the fastest, with a minimum loss function value of 1.02, which was 1.53 and 3.08 lower than that of hierarchical distance-gene expression programming and ordinary gene expression programming, respectively. When the stirrups were rusted, the predicted values were within an error range of ±20%. The average ratio between the inclination test value and the corresponding calculated result was 0.927. The predicted values were too large, and the model did not fully reflect the effects such as weakened concrete restraint caused by rust. The improved algorithm had a prediction coefficient of 0.992, which was 0.087 and 0.125 higher than that of hierarchical distance-gene expression programming and ordinary gene expression programming, respectively. The performance of various indicators in the testing set was better than that in the training set, indicating its excellent generalization ability. The improved algorithm can effectively predict the shear performance of slab beams and reduce the economic losses caused by shear failure.