Modeling with Grey Prediction of Long Term CBN Grinding Performance

[[abstract]]CBN 磨削加工(Grinding process)是精密元件的重要製程之ㄧ，是利用砂輪在高速旋轉下其表面CBN 磨粒與工件表面材料間的相對運動所進行之微量的材料移除行為，能產製高精度的表面和優良的尺寸，常用於最後一道的加工程序，以確認加工精度。製程的施作具有高度的技術性，經常需要仰賴技術人員的經驗以確保加工品質，不利於製程自動化的實現，也因此成為提升製程競爭力的阻礙。若能透過監測系統的運作，預測製程品質的變化趨勢，並建議砂輪須再削銳的時機，以輔助製程管理而增進製程安排的效率，如此則可減少人力耗費，若再搭配工件自動上料系統，則必有促成製程半自動化的契機。本計畫針對CBN 磨削進行製程趨勢預測模型的建構，以能反應製程變異的聲射訊號來關聯磨後工件之表面品質，藉數據萃取所獲得之訊號特徵值，利用灰色理論之灰色預測GM(1,1)模型針對訊號特徵值進行模型的建構，用以預測CBN 磨削加工之製程趨勢，進而判定砂輪之再削銳時機。本計畫針對所需，規劃磨削實驗，包含前置實驗及CBN 磨削主體實驗，以確認訊號與加工品質間之關連，並進行相關數據之蒐集，以利模型的建構。將實驗取樣而得之數據，分別進行處理及分析以萃取其特徵值，同時並關聯製程訊號與加工後表面品質，藉由數據特徵值建構CBN 磨削製程趨勢預則模型，藉訊號與加工品質之關連作為砂輪再削銳時機之判定依據，依此二者以提供磨削預測型監測系統之運作法則的訂定依據。[[abstract]]CBN grinding is one of the key machining process for precision element. The process utilizes surface CBN grain of the wheel rotating under high speed to perform minute material removal of workpiece with relative feed motion. CBN grinding can generate surface of high precision and quality, and is usually used in the last process for precision workpiece in order to confirm the machining accuracy. The process shall be performed under highly technical skill to ensure quality, and is therefore not easily operated automatically, which becomes an obstacle for the promotion of process compatibility. If the process runs with the aid of monitoring system to predict long term quality variation for the suggestion of wheel redressing timing, it can then assist the process management to increase efficiency and reduce the requirement of manpower to handle the process. If it is further aided with the automatic loading device, CBN grinding can then manage to operate semi-automatically. This project aims at constructing a predictive model of CBN grinding quality variation. With the characteristic value extracted from acoustic emission (AE) grinding signal to correlate ground surface quality, a grey predictive model is constructed to predict CBN grinding quality variation and to suggest the appropriate wheel redressing timing. Grinding experiment, including preliminary and main experiments, shall be designed and performed in this project to confirm the relation between the AE signal and ground quality, and to collect data form model construction. This model could be an influential reference for setting up the operation algorithm for the CBN grinding monitoring system.[[note]]NSC98-2221-E327-00