ALGORITMA VARIABLE NEIGHBORHOOD DESCENT WITH FIXED THRESHOLD UNTUK KESEIMBANGAN LINTASAN PERAKITAN TWO SIDED DENGAN KRITERIA MINIMISASI JUMLAH STASIUN KERJA

Sari

ABSTRAK

Lintasan two-sided adalah dua lintasan seri (bagian sisi kiri dan bagian sisi kanan) yang bekerja secara pararel. Pekerjaan harus dikerjakan pada bagian kiri atau pada bagian kanan dari sisi perakitan, namun ada juga pekerjaan yang dapat dikerjakan dari kedua sisinya. Algoritma Variable Neighborhood Descentwith Fixed Threshold (VND with Fixed Threshold) memungkinkan penerimaan ruang solusi lebih besar karena mempertimbangkan semua solusi yang muncul meskipun bukan solusi yang lebih baik. VND with Fixed Threshold memanfaatkan perubahan struktur yang terjadi dalam neighborhood untuk mengatasi masalah jika pencarian solusi terjebak dalam minimum lokal. VND with Fixed Threshold terdiri atas tahap pembangkitan solusi inisial  menggunakan algoritma region approach dan tahap perbaikan solusi inisial dalam pencarian local search (exchange dan insert).

Kata kunci: keseimbangan lintasan perakitan, lintasan perakitan two-sided, metaheuristik, varible neighborhood descent with fixed threshold.

 

ABSTRACT

Two-sided line is a two serial line (on the left side and the right side) working togetherin parallel. Work to be done on the left or on the right side of the assembly line, and there is also work that can be done from both sides. Variable neighborhood descent algorithm with fixed threshold (VND with Fixed Threshold) allows a larger solution space of acceptance because it considers all the solutions that emerge even though not the better solution. VND with Fixed Threshold utilize structural changes that occur in the neighborhood to address issues if the search was stuck in a local minimum solution. VND with Fixed Threshold stages consist of generating the initial solution using region approach algorithm and then improving the initial solution using local search (exchange and insert).

Kata kunci: assembly line balancing, two-sided assembly line balancing, metaheuristic, varible neighborhood descent with fixed  threshold.