Estimating punching shear capacity of steel fibre reinforced concrete slabs using sequential piecewise multiple linear regression and artificial neural network

Abstract

Estimating punching shear capacity is an important task in the design of steel fibre reinforced concrete (SFRC) flat slabs. The accuracy of commonly employed empirical design equations can potentially be improved with the use of machine learning. This study relies on a piecewise multiple linear regression (PMLR) and artificial neural network (ANN) approaches to construct a prediction model that can approximate the mapping function between the punching shear capacity of SFRC flat slabs and its influencing factors. Moreover, a sequential algorithm for automatically constructing the PMLR model structure is implemented. The algorithms of gradient descent and Levenberg-Marquardt backpropagation are employed to train the ANN based prediction models. A data set including 140 testing samples with six influencing factors of slab depth, effective depth of the slab, length or radius of the loading pad or column, compressive strength of concrete, the reinforcement ratio, and the fibre volume have been collected from the literature. This data set is then used to train and verify the sequential PMLR (SPMLR) and ANN models. Experimental results show that SPMLR can deliver prediction outcome which is better than those of ANN as well as empirical design equations. Therefore, SPMLR can be a promising alternative to assist structural engineers in the design phase of structures containing SFRC flat slabs.