Identifikasi Unsur-Unsur Berdasarkan Spektrum Emisi Menggunakan Jaringan Syaraf Tiruan

*Eko Prasetyo  -  UNDIP, Indonesia
Muchammad Azam  -  Physics Department, Faculty of Sciences and Mathematics, Diponegoro University, Indonesia
Jatmiko Endro Suseno  -  UNDIP, Indonesia
Published: .
Open Access
Citation Format:
Article Info
Section: Articles
Language: ID
Full Text:
Statistics: 434 1513

 ABSTRACT---Neural network program for elements identification based on its emission spectrum has been made using backpropagation method. The programming language which was used is MATLAB 7.0. This neural network has a single hidden layer. Training and testing data are emission spectrum data which are emission wavelength from each element. Training process was done by introducing known emission spectrum data to neural network program. Neural network program has been successful to identify elements based on its emission spectrum. Training process will be faster if we adjust the number of hidden layer’s neuron as 100, the value of learning rate as 0,049 and the value of momentum as 0,98. The neural network accuracy of identifying elements is determined by the value of error target. Error target. The value of target error about 10-2 has accuracy 97,14% and the value of target error about 10-4 has accuracy 100%.

Keywords: Neural network, backpropagation method, and emission spectrum

Article Metrics: