製造業為了因應客製化與小批量生產模式，生產現場的資訊傳遞與監控必須更為快速，使產品和服務能即時滿足顧客的需求，快速的資訊蒐集與即時監控製造狀態成為了回應顧客需求能力的關鍵。監控硬體最常見是可程式序控制器(Programmable Logic Controller；PLC)，PLC可以收集機台加工資訊，即時了解現場加工的情形，但缺點為跨平台問題，不同廠商所開發之機台內的I/O點設計皆為不同，PLC的安裝會受限於不同的機台種類與作業系統環境。本研究提出了利用電流信號為基礎的非侵入式硬體設備，蒐集機台運作時所產生的電流資料，並將這些由電流產生的資料透過嵌入式電腦上傳至伺服器後，藉由本研究提出之作業分段演算法，於電流資料取樣頻率已知且具有穩態生產與暫態生產作業的情況下，將樣本生產電流資料投入粒子群演算法求解以計算出作業分段演算法之最佳參數配置。而後以最佳參數組合之作業分段演算法辨識出機台生產作業中之暫態生產作業如機台啟動、停機、故障等情形之起始與結束時間，令使人員可辨識出實際生產之作業並用以快速計算稼動率供管理者做為決策依據。 In order to response the customization and short runs production mode, the information transfer and monitoring speed in field of production must be faster so as to meet customer needs. Nowadays, the most common monitoring hardware is Programmable Logic Controller (PLC), PLC can collect processing information in order to understand and control on-site processing. However, the main disadvantage of PLC, cross-platform issue, which means the development of machines and their I/O point design would be different according to distinct manufacturer, and it causes the limitation of PLC installation in different type of machine and operating system.In this study, we use current signal based non-intrusive load monitoring equipment to collect current signal data which is created by processing machine, and upload current signal data to server by embedded computer. Afterwards, this study input sample processing current signal into particle swarm optimization algorithm to calculate the optimized parameter configuration of the proposed segmentation algorithm. Therefore, we can use the proposed segmentation algorithm with the optimized parameter configuration to identify starting and ending time of the transient processing operation such as the starting up, shutting down and breakdown of machines from real-time current signal, thence managers can determine actual capacity and quickly calculate utilization rate as basis for decision making.
