中文摘要 本研究目的在發展精密製造系統的動態異常診斷與監控技術,主要具訊號處理、診斷技術、故障警示及避險功能。訊號處理以傅立葉與小波轉換作為穩態與暫態訊號的處理,並在轉換後擷取特徵值。診斷技術由分析系統之機構與加工特性,規劃正常與異常之製程進行實驗,由大量的實驗收集機台運作的正常與異常訊號資料,進行訊號處理分析建立關鍵診斷特徵值。同時分類異常訊號之故障類型,建立適合之異常診斷法則與警示系統。研究中主要應用對象其一為TFT-LCD微影製程中的曝光機系統,在其機構與其異常訊號分析中,找出主要誤差源為其中的遮光板機構,並分類出三大故障特徵,最後實際進行機上診斷,即時發現異常提出故障警示並排除故障。其二為精微加工動態異常監控研究,主要針對微細切削之切削顫振和斷刀之二種異常進行偵測與抑制。研究中首先由切削顫振穩定性分析中選擇顫振之切削條件進行實驗;斷刀方面以低轉速、高切削量與高進給之切削參數進行實驗,兩種異常皆進行多次實驗,由實驗資料建立顫振與斷刀監控法則。研究中以LabView撰寫智能監控系統及人機介面,最後進行實際切削監控實驗,監控系統可直接與PC-based控制器溝通,擷取加工資訊協助診斷並發出控制命令即時改變切削參數抑制顫振或緊急停機降低運作損失。
Abstract The study is to develop suitable dynamic precision manufacturing system abnormal diagnostic and monitoring technology, consisting mainly of signal processing, diagnostic techniques, fault warning and hedging function. Signal processing uses Fourier transform for the steady-state signal processing and Wavelet transform for transient signal processing, after conversing signal to extract characteristic value. A diagnostic technique is by the analysis of mechanism and manufacturing characteristics of the system, after analysis, to plan the experiment for normal and abnormal processing. It collects the normal and abnormal signal data of the machine operation by a large number of experiments, and it establishes the critical diagnostic value by signal processing analysis. At the same time, it classifies the fault type of abnormal signal, and it establishes a suitable abnormal diagnostic rules and warning system. One of the study application objects is for TFT-LCD lithography exposure system, by analysis the mechanism and abnormal signal, it finds the main sources of error which the shading plate mechanism and classifies three fault features. Finally, it diagnoses the fault online. The system detects the abnormal situation and proposes fault warning and troubleshooting timely. The other study is monitoring and control for the abnormal cutting dynamics of machine tools. It mainly detects and suppresses two abnormalities of micro cutting in cutting chatter and cutter breakage. In the study, it first proves that the rigidity of the machine tool and cutter is less prone to chatter. By the chatter stability analysis, it selects condition of cutting chatter to experiment at the same time. A cutter breakage experiment uses low-speed, high cutting volume remove and high feed parameters. There are several experiments for the two kinds of abnormities. It establishes the cutting chatter and cutter breakage monitoring rules by experimental data. Finally, it cuts actually to produce two kinds of abnormalities. at the same time, monitoring system communicates with the controller and sends control commands to the controller, in order to immediate change cutting parameters to suppress cutting chatter or emergence shutdown to reduce operating losses.