MECHANICAL PROPERTIES PREDICTION OF HEAVYWEIGHT CONCRETE USING GENERALIZED REGRESSION NEURAL NETWORK (GRNN)

摘要

In this research, from zero to 100 every 10, normal aggregates were replaced with heavy aggregate (a mixture of iron ore and barite) in the concrete. A total of 110 cylindrical specimens (15×30cm) and cubic specimens (15×15cm) were used to examine the specific gravity, compressive strength, and tensile strength of Heavyweight Concrete (HWC). The test results confirmed that by substituting Heavyweight Aggregate (HWA) iron ore and barite mixture for 10 (or higher) of regular aggregates, a specific weight greater than 2600 kg/m~3 might achieve, and the resulting product classified as HWC. In the second phase of the research, to develop the Generalized Regression Neural Network (GRNN) for estimating compressive and tensile strength, 48 data records from the specimen tests were selected randomly to find the best network with minimum mean square error (MSE) and correlation coefficient. The results confirmed that the proposed informational model could adequately estimate the mechanical properties and simplify the design processes in computational intelligence structural design platforms in the future.