結合田口方法、類神經網路、期望函數與啟發式演算法於錫膏印刷製程參數設計之最佳化

Combining Taguchi Methods, Neural Networks Desirability Function and Meta-Heuristic Algorithm for Optimizing the Parameter Design of Solder Paste Printing Process

王健鴻(Jian-Hung Wang)

指導教授 ： 蘇朝墩

國立清華大學/工學院/工業工程與工程管理學系/碩士(2013年)

摘要

隨著各類型電子產品的需求趨向於輕、薄、短、小，印刷電路板上零組件的精密度與零組件間的密度將有所增加。為了有效的控制生產成本、提升生產品質與效率，業者必須尋求更先進之生產技術來改善表面黏著技術(Surface Mount Technology, SMT)的品質。大部分的表面黏著技術問題與錫膏印刷製程有關(Pan et al. 1999 , Huang 2010, Tsai 2008)，而錫膏體積量為衡量錫膏印刷製程品質的重要指標，因此錫膏體積量的控制能力為獲得高良率和生產效益的關鍵生產技術之一。

錫膏體積量的控制屬於「作業窗」問題，且影響其產出的參數眾多，如果僅依據工程師經驗來判斷其設定值，可能導致製程不穩定而造成不良率提高。本研究針對錫膏印刷製程參數最佳化問題提出一系統性的求解程序，首先透過歷史資料與工程知識篩選出重要參數；接著使用田口方法推斷理想刮刀壓力與最佳參數水準組合。此外，本研究也提出透過倒傳遞類神經網路(Back-Propagation Neural Network)建立回應值之SN比與控制因子間的關係模式，以期望函數(Desirability Function)合併輸出值作為適應函數，再利用基因演算法(Genetic Algorithm)、粒子群演算法(Particle Swarm Optimization)、模擬退火法(Simulate Anneal)，分別求解最佳製程參數組合。本研究以一實際個案研究來說明所提出方法的有效性，透過個案研究發現本研究提出之方法可減少錫膏體積量的浪費和大幅的增加作業窗寬度以提高產品的品質。

關鍵字

表面黏著技術 ； 錫膏印刷製程 ； 田口方法 ； 作業窗 ； 倒傳遞類神經網路 ； 期望函數 ； 基因演算法 ； 粒子群演算法 ； 模擬退火法

並列摘要

With the demand for various types of electronic products tend to be light, thin, short and small. The precision of parts and the density between the parts of the printed circuit board will increase. In order to control product cost effectively and enhance production quality and efficiency, the manufacturers must find the advanced manufacturing technology to improve surface mount technology. Most surface mount technology problems are correlated to solder paste printing (Pan et al. 1999 , Huang 2010, Tsai 2008), and solder paste volume is an important quality measure of solder paste printing process. Therefore, controlling the solder paste volume is a key production technologies to obtain high–yield rate and maintain production effectiveness.

Controlling the solder paste volume is an operating window problem, plenty of parameters will affect the output; if we only rely on engineer’s experiences to determine the values, the defect rate may increase owing to the unstable manufacturing process. This study proposes a systematic procedure for parameter optimization of the solder paste printing process. First, historical data and engineering knowledge are used to determine the significant factors. Second, the ideal squeegee pressure and optimal parameter settings are determined by using Taguchi methods. Also, we propose utilizing back-propagation neural network and desirability function and integrating genetic algorithm, particle swarm optimization, and simulate anneal to obtain the optimal parameter combination. A real case was implemented and analyzed to demonstrate the proposed approach’s effectiveness. The results show that our proposed method can reduce the waste of solder paste volume and substantially increase the operation window width in order to improve product quality.

並列關鍵字

Surface Mount Technology ； Solder Paste Printing ； Taguchi Methods ； Operating Window ； Back-Propagation Neural Network ； Desirability Function ； Genetic Algorithm ； Particle Swarm Optimization ； Simulate Anneal

參考文獻

50. 蔡聰男，2003，自適應式表面黏著製程品質預測控制系統之發展，成功大學製造工程研究所博士論文。

51. 陳夢倫，2003，積層陶瓷電容印刷製程機器參數最佳化之研究，成功大學製造工程研究所碩士論文。

35. Su, C.-T., 2013, Quality engineering: off-line methods and applications, CRC Press/Taylor & Francis Group

1. Al-Refaie, A., 2009, “Optimizing SMT performance using comparisons of efficiency between different systems technique in DEA,” IEEM Transaction on Knowledge and Data Engineering, Vol. 32, No.4.

2. Chen, H.-C., Lin, J.-C., Yang, Y.-K., Tsai, C.-H., 2010, “Optimization of wire electrical discharge machining for pure tungsten using a neural network integrated simulated annealing approach, ” Expert Systems with Applications, Vol. 37, Issue 10, pp. 7147–7153.