運用人工智慧作為製造業廠區電氣設備安全診斷與風險評估工作之研究-以某公司為例

人工智慧蓬勃發展的浪潮中，人工智慧替代人類執行工作的議題被反覆提及。本研究的核心理念主要是聚焦於人工智慧如何作為「個人助理」來輔助人類執行業務，並且在現有高昂的智能系統建置成本中尋找替代的方式，人工智慧憑藉其高速運算的能力、即時回應以及強大的數據分析能力以及連結網際網路大數據的資訊運用，在輔助方面的能力還有著更多更多可以發掘的空間。本研究主要探討的價值在於將人工智慧技術的應用，從新興起的系統式軟體，高建置成本、面向企業級的解決方案，轉化成個人級別的輔助工具。 研究標的運用了電氣設備的量測數據，相對上來說電氣設備的量測數據對於公司與商業機密、個人隱私的洩漏較為無關，且數據取得相對容易，另一方面，對人工智慧進行各式與電氣設備相關安全性與風險以及災害案例的訓練，讓人工智慧的運算模式往設備的評估方向靠攏，最後在人工智慧的分析與評估完成後，透過「交叉比對與事實查核（Cross-Referencing & Fact-Checking）」的方式進行資訊的查證，確認人工智慧本身存在的「幻覺」是否出現在本研究的分析與評估中。本研究嘗試著將人工智慧視為一名職業安全衛生管理單位的執業人員，從現場交回的量測記錄與電氣設備紀錄中評估與分析電氣設備的健康與安全性，進一步建議汰換優先順序，當然就如同一般執業人員一樣，所產出的報告內容依舊需要經過各級主管的檢視與審核，但是其在輔助分析方面的能力確實具有價值。 最終本研究結果，在數據分析與安全性評估方面，簡易的生成式人工智慧足以協助。這對於缺乏專業知識和人力的中小型企業職安人員確實能夠有效的輔助提升專業能力。而作為個人助理，能有效的輔助、能解省人力與時間從而降低整體成本，能在部分功能上替代系統式的智能管理系統，正式本研究在尋求的核心價值。

Amid the booming development of artificial intelligence (AI), the topic of AI replacing human tasks has been repeatedly raised. This research focuses on how AI can serve as a "personal assistant" to assist humans in their tasks, and seeks alternatives to the currently high costs of implementing intelligent systems. With its high-speed computing power, instant response, powerful data analysis capabilities, and access to internet-connected big data, AI offers significant potential for further advancement in this area of assistance. This research explores the value of transforming AI technology from emerging system-based software, high-cost, enterprise-focused solutions, into personal assistance tools. The research utilizes measurement data from electrical equipment, which is relatively unrelated to the leakage of company secrets or personal privacy, and is also relatively easy to obtain. Furthermore, the AI was trained on various safety, risk, and disaster cases related to electrical equipment, allowing its computational model to align with equipment assessment. Finally, after the AI analysis and assessment was completed, the information was verified through cross-referencing and fact-checking to determine whether the AI's inherent "illusions" were present in the analysis and assessment. This study attempts to treat the AI as a professional in an occupational safety and health management unit, assessing and analyzing the health and safety of electrical equipment based on measurement records and electrical equipment records submitted on-site, and further recommending replacement priorities. Of course, like regular professionals, the generated reports still require review and approval by supervisors at all levels, but its ability to assist in analysis is indeed valuable.Ultimately, this study concludes that simple generative AI can be sufficient for data analysis and safety assessment. This can effectively enhance the professional capabilities of occupational safety personnel in small and medium-sized enterprises who lack specialized knowledge and manpower. As a personal assistant, it can effectively assist, save manpower and time, thereby reducing overall costs, and even replace some of the functions of a systematic intelligent management system. This is the core value sought in this research.