Analisa Prediksi Data Debit Runtun Waktu Menggunakan Jaringan Syaraf Tiruan Algoritma Backpropagation (Studi Kasus DAS Indragiri)

Abstract

Series data forecasting discharge Qn to Qn+1 using artificial neural network modeling approach backpropagation algorithm produces a good correlation between the value of the discharge predictions and actual discharge, it is also influenced by the pattern of the data available is quite good, it can be seen from the training process , testing and validation in MATLAB value that produces a fairly high correlation learning.
In the process of building a model of artificial neural networks using matlab backpropagation algorithm, obtained parameter values that provide a good learning outcome correlation, while the parameters that epoch = 2000, lr = 0.1, mc = 0.9, the parameters have been experimentally repeated on each of the various parameters that affect. variation in the data 80 (training) and 20 (test) results are slightly better correlation than the data variation 70-30 and 75-25, this is evidenced in the testing process neural network model is applied to predict discharge in 2009.
The results of this study in predicting discharge Qn+1 be developed by conducting in-depth research, so it may be a recommendation for a flood early warning system around the station AWLR locations berhala island.
Key word : Artificial neural network, Algoritm Backpropagation, forecasting