Concrete material can be expressed as a manufactured material installed by mixing specific proportions of cement, coarse aggregate such as gravel and fine aggregate such as sand, with a percentage of water. Through experiments, it has been demonstrated that by controlling some parameters of fresh concrete, such as cement grade, water/cement ratios, as well as cement dosage, water dosage and mass, within certain limits, it can improve the properties over the long term of concrete. Since current data is routinely collected and used within the framework of quality, it is noted that it is reasonable to use modern realistic data to predict the strength and durability of long-term concrete. An example of this is our laboratory work. Up to five parameters were selected for testing to represent the factors (variables) that have a direct impact on the strength of concrete. These factors are the percentage of water, the amount of cement, the amount of sand, the amount of gravel, and the coefficient of smoothness. There is a complex, non-linear relationship between these factors and the compressive strength of concrete. A typical ANN consists of three layers: the input layer, the hidden layer, and the output layer. The three-layer ANN can perform any specific function via its built-in subset. The issue here is how to estimate the concrete's pitting by other physical values that are standard in the laboratory. We introduce traditional models from the laboratory represented by the target. Based on it, we estimate the resistance value of concrete. We also do this again through the developed neural networks (LSTM) through its codes. (A. Oztas ¨ ¸, 2006) (M. Pala, 2007) (Dr. Salim T. Yousif, 2009) (Chopra, Sharma, & Kumar, 2015), (Gupta, 2013)
