MODEL PREDIKSI SLUMP BETON DENGAN ARTIFICIAL NEURAL NETWORKS- BACKPROPAGATION

STEFANUS SANTOSA Dr. Drs, M.Kom., BASUKI SETIYO BUDI S.T., M.T., JUNAIDI S.T., M.Eng., TJOKRO HADI SST., M.T.

Abstract


The design value of slump is often done manually by calculating the value of cement water factor in order to obtain the desired slump value. But these designs often unreliable. This study proposes a model prediction of concrete slump design for a variety of quality concrete with variables that are more complex than other studies. From a series of experiments with various models using Artificial Neural Network- Backpropagation (BPNN), the smallest RMSE values obtained models that can be achieved is by 0.004294661. Best Setting model parameters are Training Cycles: = 100,000, Learning Rate = 0.001, Momentum: = 0.2, Hidden Layer Size: = 10, and Number of Hidden layer: = 1.

Kata kunci : prediction, concrete slump, artificial neural network, backpropagation.


Full Text:

Untitled

Refbacks

  • There are currently no refbacks.


KETUA DEWAN PENYUNTING
JURNAL WAHANA TEKNIK SIPIL
POLITEKNIK NEGERI SEMARANG
Jl. Prof. H. Soedarto, S.H. Tembalang, Semarang,
Telp. 024-7473417, 7466420 (hunting) Fax. 024-7472396
Email : jwahana_tspolines@yahoo.com