Jurnal REKAYASA Volume 8 Nomor 1 Artikel 1
SULIANTO
(Staf Pengajar Jurusan Teknik Sipil, Fakultas Teknik, Universitas Muhammadiyah Malang. e-mail: sulianto1967@gmail.com)
ABSTRACT: River flow forecasting a year ahead is an important step in planning the operation pattern of hydraulic building which serves primarily for the supply of water. Conventional forecasting methods are widely applied today proved less satisfactory results. Application of the flow forecasting model based on artificial intelligence system developed in this study may provide better results in presenting the relationship between historical data and prediction data stream reservoir inflow Selorejo year ahead. Model-based artificial intelligence system that delivers the best performance is the system of equations developed from historical data series of monthly shifted, meaning that the flow is going to happen in the next month (t +1) will be strongly influenced by the value of the flow in the current month (t), 1 previous month (t-1), 2 months earlier (t-2) to 12 months before (t-12). Of the three types of model-based forecasting model developed by Artificial Neural Networks can show the best performance compared to other two types of models. Testing results using the ANN-based model testing data on reservoir inflow data obtained Selorejo Year 2007 performance indicator value = 0.3729 million m3/periode RMSE, MAE = 1.9124 million m3/periode, RAE = 21.79% and 6.84% RRSE. .
Keywords: flow, artificial, intelligence, forecasting, system
DOWNLOAD NASKAH LENGKAP:
ZIDDU:
SEMOGA BERMANFAAT
