OPTIMASI ARTIFICIAL NEURAL NETWORK MENGGUNAKAN GENETIC ALGORITHM UNTUK PREDIKSI MARSHALL STABILITY PADA CAMPURAN ASPAL BETON

Authors

  • Achmad Baroqah

Keywords:

Neural Network, Optimized, Genetic Algoritm, Marshall Stability, Asphalt Concrete.

Abstract

The road design should apply engineering principles to meazure trafficdensity and rapidity in minimizing crash probability. The weaknesses ofaggregat mixed aspalth concrete is causing the road quality design beingreduced.Marshall test is a technical testing to find the properness of aggregate mixed asphalt concrete in case of road design costruction.  Marshall’s stability is one of experiment result to know maximum load that  will be occupied by asphalt concrete. However, to ensure the  accuration of stability point of Marshall’s is needed a Compute Method  such Neural Network to solve diversity and unlinearity accuracy problem.  Neural Network Optimization is being tested to find the best accuration  value by using Genetic Algorithm in order to increase accuration value that  result by Neural Network. The experiment had done in obtaining optimum  architecture and increase accuracy. The result of this research that is  confusion matrix has prove Neural Network accuracy before optimized by  Genetic Algorithm is 93.83% and becomes 97.37% after being optimized.  AUC has result Neural Network is 0.975 after being optimized become  0.992. It is prove that experiment estimation of Marshall’s stability by  using Neural Network method and Genetic Alogarithm is more accurate  than Individual Neural Network Method.

Downloads

Published

2015-09-16

Issue

Section

Articles