JOURNAL

Layanan journal yang disediakan oleh Perpustakaan Universitas Gunadarma

IDENTIFIKASI FAKTOR KETIDAKNYAMANAN PADA PEKERJA PENGRAJIN PATUNG PRIMITIF MENGGUNAKAN JARINGAN SYARAF TIRUAN

Judul Artikel:IDENTIFIKASI FAKTOR KETIDAKNYAMANAN PADA PEKERJA PENGRAJIN PATUNG PRIMITIF MENGGUNAKAN JARINGAN SYARAF TIRUAN
Judul Terbitan:TEKNOIN : Jurnal Teknik Industri
ISSN:0583-8697
Bahasa:IND
Tempat Terbit:Yogyakarta
Tahun:0000
Volume:Vol. 9 Issue 3 0000
Penerbit:Fakultas Teknologi Industri Univeristas Islam Indonesia
Frekuensi Penerbitan:4X per Tahun
Penulis:Hari Purnomo, Sri Kusumadewi, Hesti Milawati
Abstraksi:A good product and performance will be influenced by operator, comfort, environmental condition and equipment. Comfort which is felt by operator will make work activitj smooth. Many factors that influence comfort are particularly related to ergonomic concept particularly. This research will try to identi fij uncomfortable factors in finishing department by artificial neural network. Backpropagation architecture and Learning Vector Quantization (LVQ) will be implemented in this research. Gradient descent is used to identify some factors that cause uncomfortable. LVQ is used to classifij some solutions to solve uncom fort. Backpropagation network has 1 input layer, 1 hidden layer, and 1 output layer. Input layer consists of 33 units, hidden layer has 20 units, and output layer has 1 unit. This network also is supported by some biases. Sigmoid activation function is used to activate units in the first hidden layer, and identity activation functions used to activate units in the output layer. This network result accuracy value is 98% . LVQ network in 'A' class (work position cause) consists of 33 units in input layer, 10 units in competitive layer, and 2 units in output layer. This network result accuracy value is 98% . LVQ network in 'B' class (repetitive/cumulative work cause) consists of 33 wilts in input layer, 7 units in competitive layer, and 3 units in output layer. This network result accuracy value is 78%. Uncomfortable causes and solutions for some different cases ivitli the same variables can be predicted by these trained networks.
Kata Kunci:backpropagation; learning vector quantization; ergonomic.
Lokasi:p. 229
Terakreditasi:belum