SIMULASI METODE BACK PROPAGATION DALAM ANALISIS HASIL PENGARUH BIJI KARET SUBSTITUSI AGREGAT KASAR TERHADAP KUAT TEKAN BETON

Abstract

Rubber seeds are one of the most quantity natural wastes in South Sumatra.  Therefore we need the processing of rubber ore waste in order to have a use value and economic value. Back propagation is one method in the use of artificial neural networks. The data generated is based on data that has been inputted with the target achievement as output data. The input data used was obtained from the results of laboratory testing of concrete compressive strength with rubber seed substitution in coarse aggregate of 5%, 10%, 15% and 20% for the life of the concrete 7 days, 14 days and 28 days. Laboratory test results show that the greater the composition of the use of rubber seeds, the compressive strength of concrete decreases. A mixture of 5% rubber seeds has the greatest compressive strength yield of 19.33 MPa. Concrete compressive strength results obtained from the test results are then used as input data in the training process with the back propagation method. The back propagation simulation method is able to predict the test results with an smallest error valued at 3.5% and the biggest error  47%. The back propagation method successfully predicts the results of a concrete compressive strength due to an error for predictions of less than 50%. This is because the input data entered is very small. Therefore, further research is needed to get maximum results.