# Levy Flightを用いたEAによる施設レイアウト問題の解法Solving Facility Layout Problems Using EA with Levy Flight

## 抄録

The facility layout problem (FLP) is a configuration problem where the goal is to determine the most efficient arrangement of interacting departments in a facility so as to optimize the working time or material cost. The departments are usually represented as rectangle objects which location and dimensions that should be optimized. Since the problem is quite complex and very difficult to solve, approximate approaches for this problem have been popular in recent years.In this paper, our goal is to determine the positions and aspect ratios of a number of irregular rectangles so that the transportation cost is minimized. We propose a method using the evolutionary algorithm (EA) based on Levy Flights (LFs) for this problem. The proposed method uses LF to search a new solution on the search space. Recently, methods based on LF have also been shown to be more effective than the based on Random Walk. Since the main operation of EA is mutation, we expect an EA-LF combination to perform well. Experiments on the benchmark problems with the number of departments from 7 to 62 indicate that the proposed method is effective for large scale problems than conventional methods such as genetic algorithms and ant colony optimization.

The facility layout problem (FLP) is a configuration problem where the goal is to determine the most efficient arrangement of interacting departments in a facility so as to optimize the working time or material cost. The departments are usually represented as rectangle objects which location and dimensions that should be optimized. Since the problem is quite complex and very difficult to solve, approximate approaches for this problem have been popular in recent years.In this paper, our goal is to determine the positions and aspect ratios of a number of irregular rectangles so that the transportation cost is minimized. We propose a method using the evolutionary algorithm (EA) based on Levy Flights (LFs) for this problem. The proposed method uses LF to search a new solution on the search space. Recently, methods based on LF have also been shown to be more effective than the based on Random Walk. Since the main operation of EA is mutation, we expect an EA-LF combination to perform well. Experiments on the benchmark problems with the number of departments from 7 to 62 indicate that the proposed method is effective for large scale problems than conventional methods such as genetic algorithms and ant colony optimization.

## 収録刊行物

• 進化計算学会論文誌

進化計算学会論文誌 9(1), 1-9, 2018

進化計算学会

## 各種コード

• NII論文ID(NAID)
130006712552
• 本文言語コード
JPN
• データ提供元
J-STAGE

ページトップへ