A hybrid data-driven optimization and decision-making approach for a digital twin environment: Towards customizing production platforms

Abstract

In the Industry 4.0 era, advanced technologies are transforming manufacturing processes and systems. Additionally, the increasing prevalence of big data and AI technologies have made decision-making using manufacturing data increasingly important. However, Small and Medium-sized Enterprises (SMEs) have encountered significant obstacles in adopting these technologies due to resource limitations and constraints. For SMEs, selecting an appropriate production strategy is challenging due to the complexity of manufacturing systems. As a response, this paper proposes a hybrid Simulation-Optimization with Multi-Criteria Decision-Making (SOMCDM) framework for SMEs to identify effective and customized production layouts. In the proposed approach, we model various production scenarios using a cellular manufacturing system. Surrogate models for different production layouts are created to basis functions using Multivariate Adaptive Regression Splines (MARS). Subsequently, the basis functions are used as fitness functions to identify optimal production parameters in a genetic algorithm. Then, optimized parameters are applied to production criteria and ranked using a multi-criteria decision-making technique. In a case study, the proposed framework is applied to select the best production platform among three scenarios for a company assembling complex products. The selected production platform improves overall manufacturing performance by 11.95% compared to the existing one. This study demonstrates the effectiveness of the proposed framework in identifying the best production platform for labor-intensive SMEs manufacturing high-mix, low-volume products using SOMCDM for a digital twin environment. The proposed framework is further detailed through a case study of a 3D printer assembly factory.