Work

Multi-objective multi-product process planning with reconfigurable machines: Exact and metaheuristic approaches

Year: 2025

Type: article

Abstract: • Proposing a new method based on the Normal Boundary Intersection employing a function to update β values for solving the Multi-Product process planning (MPPP) problem with RMTs. • Implementing Two Metaheuristics to solve the problem. • Investigating a special case of the MPPP, involving identical products: the Multi-Unit Process Planning, MUPP and retrieving some insights on solution structure. Process planning in reconfigurable manufacturing systems usually considers a single product, this reduces the efficiency of the overall production plan when multiple products are combined. This paper tackles the Multi-Product Process Planning Problem (MPPP), optimizing both individual process plans and their sequencing. We propose a 0–1 LP model, the model is relaxed by fixing the product sequencing variables and implemented in a Normal-Boundary Intersection method (NBI-es), the method uses a function for iteratively updating β values. Three metaheuristics are also developed: NSGA-II, and two variants of MOEA/D, one enhanced by Opposition-based learning (OBL). Computational experiments show that the update function enhances the performance of NBI over simple Normalized-Weighted Sum (NWS) method. Additionally, NBI-es performs better in HV metric for small size instances if it is given enough CPU times, while MOEA/D significantly outperforms NSGA-II on larger instances on most convergence and spread based metrics. OBL further enhances solution diversity for MOEA/D, albeit with less convergence. A special case of the MPPP is investigated, involving identical products: the Multi-Unit Process Planning (MUPP). An integrated approach was compared with a sequential separated approach. Results indicate that the integrated approach outperforms the separated method for smaller problem instances. Moreover, the analysis of high-quality MUPP solutions revealed a tendency towards diverse process plan combinations rather than repetitive identical ones.

Source: Expert Systems with Applications

DOI: 10.1016/j.eswa.2025.129591

Authors: Abdelkader Mechaacha, Fayçal Belkaid, Nadjib Brahimi

Institutions: University of Abou Bekr Belkaïd, École Supérieure de Commerce de Rennes

Language: English

Topic: Scheduling and Optimization Algorithms

Subfield: Industrial and Manufacturing Engineering

Field: Engineering

Domain: Physical Sciences

SDG: Industry, innovation and infrastructure

Open Access status: hybrid

Estimated APC paid: $3,500