@article{Zhang2026, 
author = {Changtian Zhang and Qi Zhang and Shujin Qin and Xiwang Guo and Bin Hu and Wenjie Luo},
title = {Multi-Objective Optimization of Defective Multi-Inventory Mother-Plate Cutting},
year = {2026},
journal = {Computers, Materials & Continua},
volume = {88},
number = {1},
keywords = {Defective plate cutting, multi-objective optimization, metaheuristic optimization, Large-Language-Model (LLM)-assisted decision support, Pareto-based search},
url = {https://www.sciopen.com/article/10.32604/cmc.2026.076620},
doi = {10.32604/cmc.2026.076620},
abstract = {The increasing complexity of steel manufacturing and the rising demand for customized high-grade plates have intensified the need for efficient and defect-aware cutting optimization. In practical production, mother plates frequently contain multiple surface defects, and the cutting process is further constrained by delay-sensitive operations such as tool-change sequences and defect-tolerance requirements. To address these challenges, this study formulates the Defective Multi-Inventory Mother-Plate Two-Dimensional Cutting Stock Problem (DMMP-2CSP) as a multi-objective model that simultaneously maximizes cutting profit and minimizes tool changes under strict geometric and defect-avoidance constraints. We develop an Improved Multi-Objective Grey Wolf Optimizer (IMOGWO) featuring continuous random-keys encoding with hierarchical decoding to handle multi-plate, multi-defect layouts; a Large-Language-Model-guided Fourth-Leader Boost mechanism that adaptively mitigates stagnation through domain-informed auxiliary-leader generation; and an NSGA-II fusion module incorporating non-dominated sorting, crowding-distance control, and stochastic variation to balance exploration and exploitation throughout the search. Extensive experiments on industrial-scale datasets demonstrate that IMOGWO consistently produces well-distributed Pareto-optimal solutions, significantly improves cutting profit, reduces tool-change frequency, and achieves superior overall performance compared with classical Multiobjective Grey Wolf Optimizer, Multiobjective Particle Swarm Optimization, Multi-Objective Cuckoo Search, and Multi-Objective Snake Optimizer baselines.}
}