A Hybrid Chaotic Gravitational Search Algorithm Based on Gower's Similarity Coefficient for Cellular Layout Design in Mass Customization Systems with Mixed-Type Data

Abstract

This research presents a novel hybrid methodology for designing cellular layouts in mass customization systems. The proposed method requires only production routing information and consists of three components: (1) Gower's similarity coefficient for mixed-type data, (2) gravity-based clustering for initial cell formation, and (3) a Chaotic K-best Gravitational Search Algorithm (CKGSA) for final layout optimization.

A hypothetical example with several products and machines was considered for validation. Results showed that Gower's coefficient successfully identified main machine clusters. The CKGSA significantly improved the fitness function, achieving zero inter-cell movement and considerably increasing intra-cell similarity. A systematic literature review confirmed that the combination of Gower's coefficient, gravity approach, and chaotic gravitational search is unique in the literature.

The main novelty lies at three levels: first application of Gower's coefficient in cellular layout problems, development of gravity-based clustering for virtual coordinate assignment, and presentation of an improved chaotic version of the gravitational search algorithm. By significantly reducing data requirements, the proposed method enables small and medium enterprises to implement cellular manufacturing systems quickly and cost-effectively.