Prediction of applied load in reinforced concrete slabs using natural frequency combined with deep learning model

Abstract

In recent decades, the application of artificial intelligence in engineering problems has become urgent. For civil structures, the prediction of the applied loads acting during their operation has practical significance. This helps to monitor the health and ensure the safety of the structure. This requirement has promoted the development of non-destructive prediction methods using vibration characteristics combined with artificial intelligence algorithms. This paper proposes a method to predict the applied load on reinforced concrete slab structures using a machine learning model that combines a deep learning model with the natural frequency. The natural frequencies of the first four vibration modes and the corresponding applied loads on the slab are used to train an artificial neural network (ANN). In order to verify the method's effectiveness, a reinforced concrete slab subjected to uniformly distributed loads is simulated by using the finite element method. The simulation takes into account the nonlinear behavior of concrete and reinforcements. The slab is loaded from zero to failure. For each load level, the natural frequencies of the first four vibration modes are extracted to train the ANN model, which consists of three network layers. The analysis results show that the first four vibration modes accurately predict the load applied to the slab, particularly at load levels where the cracks occurred.