A Hybrid Coral Reefs Optimization—Variable Neighborhood Search Approach for the Unequal Area Facility Layout Problem

L Garcia-Hernandez,A Abraham,C Carmona-Munoz,L Salas-Morera,S Salcedo-Sanz

doi:10.1109/access.2020.3010577

Abstract

The Unequal Area Facility Layout Problem (UA-FLP) is a relevant optimization problem related to industrial design, that deals with obtaining the most effective allocation of facilities, that make up the rectangular manufacturing plant layout. The UA-FLP is known to be a hard optimization problem, where meta-heuristic approaches are a good option to obtain competitive solutions. Many of these computational approaches, however, usually fall into local optima, and suffer from lack of diversity in their population, mainly due to the huge search spaces and hard fitness landscapes produced by the traditional representation of UA-FLP. To solve these issues, in this paper we propose a novel hybrid meta-heuristic approach, which combines a Coral Reefs Optimization algorithm (CRO) with a Variable Neighborhood Search (VNS) and a new representation for the problem, called Relaxed Flexible Bay Structure (RFBS), which simplifies the encoding and makes its fitness landscape more affordable. Thus, the use of VNS allows more intensive exploitation of the searching space with an affordable computational cost, as well as the RFBS allows better management of the free space into the plant layout. This combined strategy has been tested over a set of UA-FLP instances of different sizes, which have been previously tackled in the literature with alternative meta-heuristics. The tests results show very good performance in all cases.

Highlights

The Unequal Area Facility Layout Problem (UA-FLP), deals with the arrangement of spaces, machinery, or any kind of facilities in a limited area with known dimensions, and complying with a set of requirements or constraints [1]
The basic procedure is the same as in the Coral Reefs Optimization algorithm (CRO) algorithm [57], [58], but we introduce a larvae optimization with the search method proposed that will be applied when a larva gets to settle on the reef
All of them have been previously solved using FBS as layout representation. These UA-FLPs are of different size and characteristics, in order to cover the entire spectrum of different possible alternatives

Summary

INTRODUCTION

The Unequal Area Facility Layout Problem (UA-FLP), deals with the arrangement of spaces, machinery, or any kind of facilities in a limited area with known dimensions, and complying with a set of requirements or constraints [1]. Examples of them are ‘‘automated layout design program’’ (ALDEP) [18], ‘‘computerized relationship layout planning’’ (CORELAP) [19]; and ‘‘programming layout analysis and evaluation technique’’ (PLANET) [20] All these algorithms have an issue in common: It is difficult to obtain optimal solutions due to only one solution is proposed in each execution [21], so heuristic and meta-heuristic methods with population of solutions took place among the most popular approaches. On the contrary to other meta-heuristics, VNS explores increasingly distant neighborhoods of the current solution and jumps to a new one if there is an improvement, allowing a more intensive exploration of the search space [39] In this way, it keeps favorable characteristics of the current solution and obtain promising neighboring solutions. The CRO has been recently applied to the UA-FLP successfully, improving most of the previously known results by means of combining the CRO with island evolution [52] and multiobjective interactive evolution [53]

SPECIFIC UA-FLP FORMULATION

CONTRIBUTIONS OF THIS WORK

STRUCTURE OF THE PAPER The rest of the paper is structured as follows

PROPOSED APPROACH

CONCLUSIONS