Prediction of the load-bearing capacity of reinforced concrete beams with a rectangular cross-section using the basic principles of machine learning

Abstract

A step-by-step implementation of a machine learning algorithm for estimating the capacity of rectangular sections of reinforced concrete beams is considered. In this case, prestressing is not taken into account. Dependencies for strength determination based on analytical models are given, as well as the solution to the linear regression equation. The minimisation of the MSE between the data obtained from the linear regression equation and the analytical model is used as a metric to assess the quality of the predictions. A preliminary prediction of the ultimate moment is given in the case of considering a single working rebar and the plastic nature of normal section failure. The approach presented has prospects for use in the study of the load-bearing capacity of steel structures. For example, in stochastic optimisation algorithms, technical condition assessment and damage propagation prediction, structural investigation of accident causes, load identification, etc.