CD Skripsi
Analisis Penelusuran Banjir Menggunakan Jaringan Saraf Tiruan Pada Sungai Indragiri (Studi Kasus Ruas Pulau Berhalo – Peranap)
ABSTRACT
Flood routing in the Riau region is influenced by one factor, namely the
water level in the river area, water level data processing using the Artificial
Neural Network program with Back propagation Method. Data processing of
water level carried out is to predict the water level in the downstream by using the
water level data in the upstream and the water level data in the downstream.
Backpropagation method is able to produce data processing that correlates
between input data and target data. This can be seen during the training, testing
and data validation process. Design or modeling of artificial neural networks
using the MATLAB program, while the parameters used are Epoch, lr or learning
rate, and the value of momentum or mc. The data used are 2550 water level data
from year 2010-2016 with grouping of data to 70% as training data and 30% as
testing data and validation data as much as 100%, in testing this application the
prediction results obtained are less good because RMSE values which is too large
so with that transformation of data into the form of logarithms, data processing
results of logarithms can reduce the value of RMSE which is pretty good but the
correlation between upstream and downstream data is very significant so that it
affects the prediction results.
Keywords: Artificial Neural Networks, Backpropagation, MATLAB, Epoch,
Learning rate, Momentum, RMSE, water level prediction
Tidak tersedia versi lain