Evaluation of Relationship between Cement Composition and Abrasion Resistance by Using Artificial Neural Network

Abstract

The abrasion resistance of a concrete and/or cement mortar is changing depending on some properties like compressive strength, matrix structure, gap ratio, aggregate type. On the other hand cement composition has also an effect on this resistance. Numerical calculation methods, originated as parallel with the developments on today’s computer technology, provide a great advantage especially in prediction of experimental results. Thus, in this study, it is aimed to investigate the relationship between abrasion resistance and cement composition by using standard test methods together with predictions of Artificial Neural Networks (ANN), that is a sub-branch of Artificial Intelligence (AI). When modeling ANN, it is benefited from multi-layer ANN, which uses supervised learning rule. During training and testing stage of ANN, the results obtained from Bohme abrasion tests of cement samples having seven different compositions are used. For these samples’ curing time are chosen as 7, 28, 90, 180, 270 and 360 days. By testing of ANN on conclusions, its usability, advantages and disadvantages are evaluated.