(528i) Advancing Manufacturing Sustainability Research: A Domain-Specific Large Language Model Perspective

Advancing research in manufacturing sustainability requires the development of more comprehensive and broadly applicable sustainability metrics, enhanced assessment methods, and robust decision-making tools capable of functioning under uncertainty. These innovations must be built upon thorough reviews of existing methodologies, technologies, tools, and well-documented industrial case studies. This is a particularly complex task, as some critical data and knowledge are publicly accessible, while others remain confidential or restricted.

One promising solution to this challenge is the use of large language models (LLMs). However, general-purpose LLMs may lack the depth required for highly specialized industrial applications, domain-specific LLMs are often necessary. These models should be effective tools for helping pursue manufacturing sustainability, by enabling smarter, faster, and enhanced sustainability assessment, system analysis, informed decision-making, and effective communication tailored to sustainability contexts. Specifically, fine-tuned domain-specific LLMs can compare and recommend sustainability indicators, identify the minimum set of system parameters required for evaluating those indicators, and suggest the most appropriate assessment methodologies. If the dynamic sustainability is aimed, real-time monitoring is essential. In this context, domain-specific LLMs can extract and summarize relevant performance data from production logs, reports, and Industrial Internet of Things (IIoT) systems, facilitating continuous improvement.

In this paper, we present a general framework for developing domain-specific LLMs tailored to manufacturing sustainability. A case study is also provided to illustrate the potential benefits of this approach for advancing research in sustainable manufacturing.