Claim Missing Document
Check
Articles

Found 2 Documents
Search

Model Jaringan Syaraf Tiruan untuk Prakiraan Harga Komponen Bahan Baku Pakan Unggas di PT XYZ Ahmad Haris Hasanuddin Slamet; Bambang Herry Purnomo; Dedy Wirawan Soedibyo
Industria: Jurnal Teknologi dan Manajemen Agroindustri Vol 9, No 2 (2020)
Publisher : Department of Agro-industrial Technology, University of Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21776/ub.industria.2020.009.02.9

Abstract

AbstrakPT XYZ adalah salah satu produsen pakan unggas di Kabupaten Banyuwangi, Jawa Timur. Permasalahan dalam pengembangan pakan unggas di PT XYZ adalah harga pakan unggas yang berfluktuasi. Komponen terbesar bahan baku pembuatan pakan unggas adalah jagung dan bungkil kacang kedelai (BKK). Permasalahan fluktuasi harga pakan unggas dapat diatasi dengan prakiraan harga jagung dan BKK. Prakiraan yang tepat dapat membantu PT XYZ untuk optimalisasi alokasi sumber daya perusahaan. Optimalisasi sumber daya bertujuan untuk meningkatkan keuntungan yang diperoleh perusahaan. Tujuan dari penelitian ini adalah pengembangan model jaringan syaraf tiruan (JST) backpropagation untuk prakiraan harga jagung dan BKK. Model JST dikembangkan dengan perlakuan jumlah lapisan tersembunyi (node hidden layer), fungsi aktivasi, dan laju pembelajaran (learning rate). Data penelitian yang digunakan adalah harga jagung dan BKK pada periode Januari 2016-Oktober 2018. Hasil penelitian menunjukkan bahwa model JST terbaik untuk prakiraan harga jagung adalah 12 node input, 5 node hidden layer, dan 1 node output dengan kombinasi fungsi aktivasi sigmoid biner (logsig)-sigmoid biner (logsig) dan learning rate 0,005. Model JST terbaik untuk prakiraan harga BKK adalah 12 node input, 10 node hidden layer, dan 1 node output dengan kombinasi fungsi aktivasi sigmoid bipolar (tansig)-pure linier (purelin) dan tingkat learning rate 0,006.Kata kunci: harga jagung dan bungkil kacang kedelai, Jaringan Syaraf Tiruan, pakan unggas AbstractPT XYZ is one of the poultry feed producers in Banyuwangi Regency. The problem in developing poultry feed at PT XYZ was related to the fluctuative price of poultry feed itself. The biggest component of raw material for producing poultry feed that affect prices were maize and soybean meal. The problem of poultry feed price fluctuations can be overcome by forecasting the price of maize and soybean meal. The accurate forecast can be used as a reference for PT XYZ in optimizing the allocation of resources so as to increase the profits of the company. The aim of this study was developing a backpropagation neural network (ANN) model. The ANN model was developed by number of hidden layers, activation function, and learning rate. The price of maize and soybean meal in the period January 2016-October 2018 was used as data in this study. The best model for forecasting maize price was 12 input nodes, 5 hidden layer nodes, and 1 output node with a combination of the sigmoid binary (logsig)-sigmoid binary (logsig) activation function and 0.005 learning rate. The best model for forecasting soybean meal was 12 input nodes, 10 hidden layer nodes, and 1 output node with a combination of sigmoid bipolar (tansig)-pure linear activation function (purelin) and 0.006 learning rate.Keywords: Artificial Neural Network, maize and soybean meal prize, poultry feed
PRAKIRAAN HARGA MEAT BONE MEAL (MBM) MENGGUNAKAN JARINGAN SYARAF TIRUAN BACKPROPAGATION Ahmad Haris Hasanuddin Slamet; Bambang Herry Purnomo; Dedy Wirawan Soedibyo
JURNAL AGRIBISNIS Vol. 10 No. 1 (2021): Mei 2021
Publisher : Program Studi Agribisnis, Fakultas Pertanian

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32520/agribisnis.v10i1.926

Abstract

XYZ is a poultry feed producer in Banyuwangi Regency, East Java. The problem in developing poultry feed at PT XYZ was the fluctuating price of poultry feed. Meat bone meal (MBM) or what is called meat flour is one of the raw materials for poultry feed that affects the final price of poultry feed products. The price of MBM was greatly influenced by the exchange rate of the rupiah against the dollar. Forecasting is one way that needs to be done in dealing with MBM price fluctuations. The aim of this study was to estimate the price of MBM using backpropagation neural networks (BNN). The data used in this study was the price of MBM in the period January 2016-October 2018. Based on the results of the study, the best BNN architecture for the estimated MBM price was12-10-1 (12 input nodes, 10 hidden nodes, and 1 output node). This architecture has reached the training target of 0.002 with a MAPE test value of 13.93%. Based on forecasts with the BNN the highest MBM price in May 2019 and the lowest MBM price in January 2019.