A comparative study of hybrid estimation distribution algorithms in solving the facility layout problem

The Estimation Distribution Algorithm (EDA) is an evolutionary algorithm that uses probabilistic models to create candidate solutions. Previous researchers have suggested various hybrid methods to avoid the premature convergence of EDA. This research conducts a comparative study between several variations of hybridization in EDA with regards to the descriptive statistics in the objective values. This study also proposes a new hybrid approach, named Adapted EDA (AEDA), by adapting the structure of EDA by adding a lottery procedure, an elitism strategy, and a neighborhood search. The proposed AEDA, several hybridizations of EDA, and Genetic Algorithm (GA) plus Tabu Search (TS) are applied to the facility layout design in manufacture – Enhanced Facility Layout Problem (EFLP) – to analyze their solutions. The hybrid EDAs that are being compared are EDA plus GA (EDAGA), EDA plus Particle Swarm Optimization (EDAPSO), the combination of EDAPSO plus TS (EDAhybrid), and AEDA. The experimental results show that the AEDA can significantly improves the solution quality in solving all the EFLP instances compared to other algorithms.