Optimizing Production Planning for Small Textile Enterprises with LLMs
PPCP; LLMs; Artificial Intelligence; SMEs; Textile Industry.
The textile and apparel industry has economic relevance in the state of Rio Grande do Norte and is predominantly composed of micro and small enterprises. In this context, empirical practices still prevail in production planning and control, combined with low levels of adoption of decision-support technologies. This dissertation aims to develop and evaluate a decision-support approach for Production Planning and Control in micro and small textile enterprises, based on the use of large language models. The research adopts an applied and experimental approach with a mixed design. Real production data from workshops participating in the Mais Performance Moda program were used. The proposed solution integrates classical prioritization heuristics, such as Shortest Processing Time and Earliest Due Date, with a language model configured as a Production Planning and Control assistant, capable of interpreting technical sheets in PDF format, generating resource allocation plans and simulating dynamic workforce reallocations. Results obtained in a controlled environment indicate reductions in makespan and order lead time, as well as improvements in workload balancing and operational efficiency. The solution presented response times compatible with production replanning in small-scale enterprises. From a scientific perspective, this study contributes by exploring the use of large language models in production scheduling problems using real industrial data. From a practical perspective, the approach provides a structured decision-support mechanism for managers of micro and small textile enterprises.