- 學位論文
應用人工智慧方法與田口方法優化製程參數－以射出成型機為例
Application of Artificial Intelligence Method and Taguchi Method for Parameter Optimization : Case Study in Injection Molding Machine
摘要
在競爭激烈的世代，能夠降低成本、提高品質、提高產能一直是大家所追求的，尤其是在傳統產業上。本研究透過品質工程裡的計數型田口方法，並且建立一個類神經模型來預測不良率，透過射出成型製程的個案，來證實本研究所發展的模式能夠有效降低不良率，透過計數型田口方法得出最佳的參數組合，其中包含透過各缺點對品質的影響程度進行分類的累計分析法，以及將百分比等資料型態進行轉換成可加性數值的ꭥ轉換，透過以上兩種方式來有效的挑選出影響重要製程的參數。據結果顯示，硬碟外殼的不良率由6.0%降為3.0%，改善50.0%的幅度，證明實驗的有效性與可行性。最後利用多層感知器(Multilayer Perceptron)建立一個預測模型，減少個案公司進行不必要的實驗。
並列摘要
In the fierce competition generation. Can reduce the cost, improve the quality, improve the performance has always been the pursuit of everyone, especially in the traditional industrial system. In this study, the attribute Taguchi method in quality engineering and set up a kind of model to detect the defective rate. Through the injection molding process of the case, to prove that the development of the model can effectively reduce the defective rate. Through the counting type of Taguchi method, the best combination of parameters is obtained, including the cumulative analysis method that classifies the influence degree of each defect on the quality, and the conversion of the data type such as the percentage into the additivity value. There are two ways to effectively select the parameters that affect important processes. According to the results, the defective rate of shell outside the hard disk was reduced from 6.0% to 3.0%, and the amplitude of 50.0% was improved to prove the effectiveness and feasibility of the experiment. Finally, multi-layer perceptron (Multilayer Perceptron) is used to build a predictive model to reduce the need for case companies to conduct unnecessary experiments.