:::

臺灣能源期刊論文全文

臺灣能源期刊第12卷第2期內容

出刊日期:June, 2025

題目
探索大型語言模型在工廠節能中的潛力
Title
Exploring the Potential of LLMs for Energy Conservation in Factory Setting
作者
呂宥陞、劉彥伯、劉子吉
Authors
Yu Sheng Lu, Yen Po Liu, Liu Tzu Chi
摘要
截至目前,大型語言模型(LLM)已被應用於多個行業,因為它有助於加速各行各業的工作流 程,並提供不同的功能。透過技術如檢索增強生成(Retrieving Augmented Generation)、提示工程 (Prompt Engineering)及安全防範機制(Guardrail mechanisms),這些技術為LLM在工廠環境中節能創 造了機會。LLM具備透過對話式介面與使用者互動的優勢,能夠提供直觀且自然的操作體驗。由於 其內建專業知識,LLM能協助使用者理解複雜資訊,有效縮短技術人員與非技術人員之間的知識 差距。這不僅有助於新進員工的培訓,也讓管理階層能更快速地取得現場資訊。相較之下,傳統的 能源管理系統(EMS)通常需要使用者具備一定的技術背景,並花費時間操作系統以取得所需資料, 導致管理人員在資料分析與決策上效率較低。能源優化在節能及降低工業運營成本上能帶來巨大影 響。LLM可以通過分析歷史使用數據、生產計劃及其他外部因素,如天氣,來支援能源管理。本研 究探討了LLM在工廠節能中的應用,並強調了當前的應用情況、挑戰及未來的研究方向,旨在提升 工業設置中的效率。
關鍵字
大型語言模型(LLM)、檢索增強生成(RAG)、提示工程(Prompt- Engineering)、 安全防範機制(Guardrail)、節能 收
Abatract
As of right now Large Language Model (LLM) has been used in many different industries, since it helped speed up work process in different industries with different functionality. With techniques such as, Retrieval Augmented Generation (RAG), Prompt Engineering, and Guardrail mechanisms, these opens up opportunities for LLM to conserve energy in factory setting. LLM offer a key advantage through their conversational interface, enabling intuitive interactions with users. Since the LLM is equipped with expertlevel knowledge, it can assist users in understanding complex information, thereby bridging the gap between technical and non-technical employees. This facilitates the training of new staff and allows management to access on-site information more quickly. In contrast, traditional Energy Management Systems (EMS) often require users to navigate complex interfaces and possess prior technical knowledge, which slows down data retrieval and decision-making processes. Energy optimization can make huge difference on energy conservation and save money in industrial operation. LLMs can support energy management by analyzing historical usage, production schedules, and other external factors such as weather. This paper explores the integration of techniques that can be utilized by LLM in factory energy conservation, highlighting current application, challenges, and future research directions on enhanced efficiency in industrial setting.
Keywords
Large Language Model (LLM), Retrieval Augmented Generation (RAG), Industrial, Energy Conservation, Agent.