Predição de resistência de concreto de alto desempenho através de redes neurais artificiais

Abstract

Although it is perhaps the most popular material in civil construction, concrete requires a lot of attention in its dosage (especially high-performance concrete), and often requires numerous empirical adjustments to ensure compressive strength compatible with the designed. Considering that most of the consolidated methods for concrete dosage uses abacuses and empirical tables, it is noted, in this absence of a direct mathematical method, a promising possibility of applying Artificial Neural Networks to have a simpler and reliable method for concrete dosage and determination of high-performance concrete compressive strength based on the proportion of the materials that compose it, since they are computational models very effective in solving complex problems. This work aims to develop a model to predict the compressive strength of high-performance concretes through the use of Artificial Neural Networks based on the proportions of the materials that constitute the concrete, obtaining remarkably satisfactory results.