Neural-Based Annealing Processes Optimization of Cold-Rolling Coil by Genetic Algorithm and Grey Relational Analysis

用基因演算與灰色關聯分析法於冷軋鋼捲退火製程神經網路式最佳化之研究

張士行(Shin-Hsing Chang) ； 張峰瑞(Spencer Chang) ； 張欽華(Algo Chang)

《中華技術學院學報》 24期 (2002/03) Pp. 117-126

https://doi.org/10.7095/JCIT.200203.0117

摘要

冷軋鋼在國內用途廣泛，可製成一般金屬家具外，亦可再加工，製成電腦外殼，自行車零組件等，因此我們以冷軋鋼中之CQl（一般商用品級）材質冷軋鋼捲為材料。我們考慮軋下率、第一段恒溫溫度、第二段恒溫溫度、第一段恒溫時間、第二段恒溫時間，等五項為評量指標，求取出最佳製程條件，在調整其相關退火製程參數時，需要進行一連串試誤法，常需耗時且不易得到最佳的設計，採用灰色關聯分析法，可在最少實驗次數下求得最佳化的製程。

關鍵字

信度效度分析 ； 直交表 ； 變異數分析 ； 灰關聯分析 ； 類神經網路製程最佳化 ； 基因演算法

並列摘要

Cold rolling coil annealing processes is a very widely used technology in the automobile body industry, yet the characteristics of its product are easily affected by the processing conditions such as ratio of rolling, first stage isothermal temperature, second stage isothermal temperature, time of first stage isothermal temperature, time of second stage isothermal temperature. In this paper, an effective methodology for cold rolling coil annealing processes optimization is proposed. The method includes validity and reliability analysis of data, orthogonal array of experimental design of Taguchi, analysis of variance (ANOVA), neural-based manufacturing process optimization, Genetic Algorithm and grey relational analysis.By using these methods, we can find the optimal annealing process easier and faster than the traditional method that try and error in higher strength, higher stiffness, and faster productivity and safety materials to meet the demands upon structure design and economic benefit.

並列關鍵字

Validity and reliability analysis ； Orthogonal array ； ANOVA ； Grey relational analysis ； Neural-based manufacturing process optimization ； Genetic algorithm and grey relational analysis