PEMODELAN DESAIN CAMPURAN BETON DENGAN BACKPROPAGATION NEURAL NETWORKS
Concrete is a mixture of materials that has complex characteristic so that it raises a variety of very complex models as well. The experts in concrete mixing believe that the formula to find compressive strength of a mixture is not good enough. Every mixture design only applies to one mixture only. Because of that, every mixture production who need even the slightiest diferrences in the base materials, will need a new mixture design. Concrete mixture modeling process is done manually with a variety of mixed composition and destructively testing has some drawbacks like expensive, unpredictable, and not environmental friendly. Besides of that, state of the art concrete mixture design modelling computation with Multilayer Perceptron Artificial Neural Network s (MLP) have RMSE = 5,27. Computational model developed in this study with the same data sets has more good performace than MLP model. From the results of experiments that have been carried out proved that the proposed model, Backpropagation Neural Network (BPNN), has lower error rate than MLP with RMSE = 4.18.