以實驗設計方法學建立線切割機切削速度之模型

Abstract

放電加工為模具工業的重要應用，加工參數設定影響放電加工後之加工品質，因此發展出一套有系統的自動化調整機制之智慧型加工系統，為放電加工業者努力的目標。要達到自動化調整機制，必須先經過一連串實驗設計，了解系統與加工參數之關係，繼而建立系統數學模式，本著所建立之系統數學模式可後續對放電加工參數做最佳化設計及適應性控制。 過去的有關工具機之加工品質的研究，鮮有學者採用以統計學為基礎的實驗設計方法學，有系統的利用實驗設計的標準步驟，探討系統有關之加工參數，循序漸進的提出一系列有系統的實驗規劃，找出重要加工參數與系統之關係，進而建立系統之數學模式。大部份的研究不外乎採用最佳猜測途徑法或一次ㄧ因子法。田口式的實驗規劃法簡單、易學、容易操作，近年來廣泛的被學者採用。由於其實驗次數過少，因子間的交互作用，平均分散至每個主效應中，致使所產生的混淆現象非常嚴重，致使對所得到之資訊無法做出正確的判讀，或是判讀錯誤，導致所得到的結果常常是不可採信的。 本研究擬採用實驗設計方法學，利用實驗設計的標準步驟，循序漸進的提出初步實驗、進階實驗、最終實驗，一系列有系統的實驗規劃。本研究經由一連串的實驗，其中經過對因子的範圍放大、增加實驗次數、刪除不重要之因子而投影成較少重要因子的實驗設計、摺疊設計、重複實驗等等過程，確認系統之重要因子與交互作用，最終得到系統之數學模式，並經過實務加工實驗驗證所建置之系統數學模式之準確性。

Discharge manufacture is an important appliance of the molding industry, discharging machining parameter setting influences the quality after the discharge manufacture. Hence, to develop a set of systemic automated adjustment mechanism of artificial intelligence for the discharge machining system is a mutual goal of the industry. To achieve the automated adjustment mechanism, a serious of experiments has to be performed to understand the relationship between the system and the machining parameters. A system mathematical model can then be established. Based on the established model, the cutting conditions can be optimized by the adjustment of the machining parameters. Few researchers have put their effort on the machine tool development through a statistically based experimental design algorithm to perform their experiments, find the important machining parameters, and then establish a mathematical model to describe the machine system. Many past researchers adopted a best-guess or one-factor-at-a-time approach. Taguchi experiment method is widely used because of its easy learning and user friendly interface. Due to the insufficient number of experiments, serious alias phenomenon occurs, and users would not be able to determine the important effects or/and the interactions. Thus, the model established is not reliable. This research intends to adopt the experiment design methodology, and apply its standard procedure to develop a series of systematic experiments. A preliminary, a secondary, and a final stage of experiments will be performed systematically in this study.Through a three-stage well-designed experiments, the important factors and their suitable ranges would be determined, a original fractional factorial designs can be projected into stronger designs, and a larger design can be assembled by the combination of two (or more) fractional factorial designs. Finally, the main effects and their interactions can be assured and a completely replicated full factor experiment design would be obtained. Therefore, a reliable machine system mathematical model can then be established.