加工製程資訊可視化與GAI 能耗優化技術

本研究針對高混合少量、交期緊迫及淨零碳排需求，提出結合物理建模、動態插補運動學、數位雙 生與生成式人工智慧 (GAI) 之智慧加工協作平台。系統採微服務架構，整合 NC-Code 加工時間估測、能 耗估測模型、輪廓誤差估測及切削力學分析，自動產生多組加工方案，並透過進給率與主軸轉速參數掃 描建立加工時間、能耗與輪廓誤差三維解空間。再結合 Pareto 多目標篩選進行決策分析，協助選擇代表 性的候選方案。最後將數位雙生模組封裝為 RESTful API，結合大型語言模型 (LLM) Tool Calling，自主調 用估測模型、數據分析、優化報告生成及 NC-Code 自動改寫，提升加工效率、降低能耗並兼顧加工品質。

This study proposes an intelligent machining collaboration platform that integrates physics-based modeling, dynamic interpolation kinematics, digital twin technology, and Generative Artificial Intelligence (GAI) to address the challenges of high-mix low-volume production, short delivery cycles, and net-zero carbon emissions. The platform adopts a microservice architecture and integrates NC-code machining time estimation, energy consumption estimation, contour error estimation, and cutting mechanics analysis to automatically generate multiple candidate machining strategies. Through parameter sweeps of feed rate and spindle speed, the system constructs a three-dimensional solution space comprising machining time, energy consumption, and contour error. Pareto-based multi-objective screening is then applied for decision analysis to identify representative candidate solutions. Finally, the digital twin capabilities are encapsulated as RESTful APIs and integrated with Large Language Model (LLM) Tool Calling, enabling autonomous invocation of estimation models, data analysis, optimization report generation, and automatic NC-code rewriting, thereby improving machining efficiency, reducing energy consumption, and maintaining machining quality.