Neural Network Prediction Model of Concrete Compressive Strength Size Effect

To explore the mechanism of the influence factors on the size effect of concrete compressive strength, a prediction model of concrete strength by BP neural network was established from the concrete material level in this study. An artificial neural network model was established by using seven input characteristics, namely, specimen shape, specimen section size, specimen height/diameter ratio, curing age, water-cement ratio, cement mark and maximum coarse aggregate particle size. According to the above input characteristics, the compressive strength of different section sizes was predicted, and the influence of various factors on concrete size effect was studied. Results show that the water-cement ratio has a great influence on the size effect of compressive strength of concrete with the decrease of its value, and the size effect becomes more obvious. The increase of the aspect ratio leads to the enhancement of size effect. When the aspect ratio is greater than 2, the size effect tends to be stable. The size effect becomes more and more obvious with the increase of the coarse aggregate size, however, the growth rate decreases with the increase of the coarse aggregate size. The influence of specimen shape on the size effect of compressive strength of ordinary concrete can be ignored.