Aplikasi Jaringan Syaraf Tiruan Berbasis Radial Untuk Menentukan Prediksi Waktu Pengeringan Gabah Pada Pengering Radiasi Infra Merah

Abstract

Abstract : The aim of this research is for predicting the time of drying on infrared radiation dryer by using artificial neural network with Radial Base Function (RBF) Algorithm. The RBF neural network was created and trained by using set of data that are taken from actual measurement. The parameters are used on network training namely radiation intensity on preheating and main heating stage, variety input power of the infra red on preheating and main heating stage, kinds of reflector on main heating stage and thickness of paddy for determining drying time. There were 36 variations data, includes 26 data are taken as training data for the net to learn. The rest data will be taken as test data for the net. Results showed that prediction data were closed to actual results. The MSE for the drying time prediction for training and testing average by using RBF showed 0.0003 % and 2.62% respectively compare to backpropagation 1.064% and 6.595% respectively. For the computation time showed RBF was faster than backpropagation. The time computation results for RBF and backpropagation achieved 0.511 sec and 28 sec respectively. From above result, it can be concluded that RBF is very fast and accurate for predicting drying time on infra red radiation dryer model compare to backpropagation.

Keywords : Artificial neural network, RBF, Backpropagation, Prediction, Drying time