Study on a data driven intelligent hybrid model for multi-output prediction of steel fibre reinforced concrete strength parameters

Abstract Steel fibre reinforced concrete (SFRC) is generally utilised in structural construction because of its enhanced post-cracking behaviour and mechanical properties. Classical modelling methods for SFRC behaviour tend to rely on empirical relationships, which are limited in number and sometimes may not represent intricate interactions between design variables. In order to transcend the limitations of the existing method, this paper suggests a new prediction system using a Kurková–Kolmogorov–masterpiece optimisation network (KKMO-Net) to predict SFRC strength and fracture behaviour effectively. Experimental data were obtained from extensive literature on compressive strength, flexural strength, and load-deflection behaviour. Pre-processing was conducted by normalising input parameters and removing noise, followed by the masterpiece optimisation algorithm to select the most important features that influence SFRC performance. These selected features were then passed to KKMO-Net to predict output parameters such as peak load, maximum deflection, and fracture energy. The model was implemented and validated in Python and compared to existing methods based on performance indicators like accuracy and error rate. The method suggested indicates effective and consistent forecast of the performance and can provide a powerful evaluation of the early of SFRC behaviour.