CD Skripsi
Prediksi Kuat Tekan Bata Ringan Berdasarkan Parameter Densitas Dengan Metode Jaringan Saraf Tiruan
ABSTRAK
Lightweight brick was one of new innovation in building material field using chemical foaming
agent, which reduce wight of brick significantly. Lightweight was one of advantage that
lightweight brick has, which is has significant effect in reducing structure weight when
lightweight brick apply on highrise building.Compressive strength was one of parameter that
used to measure the quality of lightweight brick, while density was one of factor that affect
compressive strength value. Artificial neural network analystic was a numeric prediction
method that replicate neural system of human body. The purpose of this research was to predict
compressive strength value using artificial neural network method.training and testing data of
ANN that used on this research was the result of eksperiment sample test, sample geometri
which used in this research was cube sampel with 10 mm side length. 120 sample was made
on eksperiment process, 105 sample used as training data and 15 sample used as testing data.
The optimal result of ANN training was ANN with 5 neuron hidden layer with 0.052336 mean
square error value, regression training as well regression testing value 00.93981 and 0.95667
each. Prediction result of ANN with 5 neuron hidden layer was 0.009176 mean square error
value and 0.98763 regression value. Base on this result can be concluded that Compressive
strength prediction of lightweight brick with ANN method base on density parameter has given
best dan accurate results.
Keyword: compressive strength, lightweight brick, artificial neural network, mean square
error, regression
Tidak tersedia versi lain