多品質數控精削之泛用模糊最適化研究

GENERAL FUZZY OPTIMIZATION ON MULTI-OBJECTIVE CNC FINISH TURNING

潘昌鑫(Chang-Hsin Pan)

指導教授 ： 王明庸

大同大學/工程學院/機械工程學系所/碩士(2008年)

摘要

多品質數控精密製程之參數最佳化問題中，常存有高度複雜的非線性關係，且控制參數與品質目標之間交互影響複雜；現有多品質之最佳化研究，皆屬特定狀態下之方案，更多需透過製程設備之大量重複操作得以達成；因多品質精密數控製程之最佳化實屬不易，本研究選定數控精密車削為題，以切削速度、進給量、切削深度、刀尖中心度為控制因子，整合模糊語意分析與理想解類似度順序偏好法(TOPSIS)，以模糊層級分析法(FAHP)探討參數控制對表面粗糙度、刀具磨耗比、材料移除率、切削噪音(推廣品質)等品質目標之影響，藉由不經設備操作，利用語意推論方式，發展一套泛用最適化機制。
本研究首先建構品質與因子間之相關性階層，透過模糊層級分析法，客觀分析各項品質權重；進而利用田口規劃 直交表，以模糊理論為基礎，依梯形隸屬函數定義，對個別品質建立語意規則，利用重心法解模糊後，探討品質之量化數值；透過語意實驗及切削實驗，獲取各項品質數據，並以理想解類似度順序偏好法整合多品質特性，獲取排序值。依據研究者翁義哲與黃得誌在81組全因子實驗所得數據進行多品質整合，進而驗證、分析、比對並確認多品質數控精密車削之「泛用模糊最適化方案」，加強模糊語意對多品質最適化方案之可行性，進而探討對推廣品質之運用價值。
由本研究結果顯示，以語意實驗所得最適切削參數排序值與實驗差異小，且品質表面粗糙度優於實驗38.7%、刀具磨耗比則為7.9%，代表最適化參數組合接近實際實驗的趨勢(排序為第8名)，也證明有很好的品質特性，符合研究預期的方向；對於推廣目標後，因增加目標也同時增加整體模糊性的範圍，使得排序值落後實驗33名，而在品質特性有良好展現。依本研究結果，以考量三項多目標能貼近實際需求，值得作為一套經濟前瞻的「多品質泛用模糊最適化」機制之參考依據，可不經實驗操作，即能有效降低成本、提昇品質，更對數控精密產業競爭力與發展有所正面激勵。

關鍵字

模糊理論 ； 模糊層級分析法 ； 切削噪音 ； 材料移除率 ； 刀具磨耗比 ； 表面粗糙度 ； 田口法 ； 理想解類似度順序偏好法

並列摘要

The issue of making parameter optimization of multi-objective CNC finish turning system regulation usually has highly complex non-linear relations, and the interactions are complex between the parameter and objective. Existing optimization research studied by a projection of the specific state, and more needed to make it by a large amount of repetitive operation and equipment. Because making parameter optimization of multi-objective CNC finish turning system regulation is really difficult, this paper chose multi-objective CNC finish turning system regulation to be topic, took cutting speed, cutting depth, feed rate and tool nose runoff as the factors of cutting, combined fuzzy linguistic analysis and “Technique for Order Preference by Similarity to Ideal Solution” (TOPSIS), and discussed the control parameter to influence of surface roughness, tool wearing, material removal rate, cutting noise (spreading objective) with FAHP. Utilize the linguistic inference way and operate without equipment to develop a set of general optimization mechanisms.
First, this research constructed the relativity stratum of quality and factor, analyzed multiple quality weight objectively by FAHP. Based on Fuzzy theory further, set up the linguistic rules for specific objective. After defuzzification by center of gravity method based on the definition of trapezoid membership function, this research got the quantization value of quality. Applied to the Taguchi orthogonal arrays further, obtained quality data by the experiment of linguistic and the one of turning. And then using TOPSIS was applied to combine multi-objective and to obtain the rank value. This research used TOPSIS to combine the general full factorial data done by Yi-Zhe Weng and Te-Chih Huang.To manifese, analyze, compare and correct the “scheme of general fuzzy optimization” on multi-objective CNC finish turning, this research strengthened the feasibility of the fuzzy linguistic analysis to the multi-objective optimization, and then discussed the application value of spreading objective.
The results shown that, the different is small between optimization parameter members of the linguistic experiment (The rank is NO.8) and turning experiment. Besides, the objective of surface roughness surpasses 38.7% and tool wear is better 7.9%. It represents close to actual experiment and proves that there are pretty good quality characteristics. Accord with and study the anticipated direction. After this section, this paper discusses popularizing quality. Because of increasing objective also simultaneously increasing the whole fuzzy scope, the rank is backwardness to 33. But still have good quality characteristics. In summary, this research presses most close to the actual demand in order to consider three objectives, and is able to be as a set of basis mechanism of economy and prospect “general optimization of multi-objective”. Without operating the equipment, this research provides not only to lower costs, to promote quality effectively, but to encourage positively competitiveness and development in CNC finish industry.

並列關鍵字

Cutting Noise ； Material Removal Rate(MMR) ； Tool Wearing Ratio ； Surface Roughness ； Fuzzy Analysis ； Taguchi Method ； FAHP ； TOPSIS

參考文獻

[1] 翁義哲，「應用模糊語意於多目標數控車削之最佳化研究」，大同大學機械工程系碩士論文， 2007。

[2] C. Y. Nian, W. H. Yang, Y. S. Tarng, “Optimization of Turning Operations with Multiple Performance Characteristics”, Journal of Materials Processing Technology, Vol.95, 1-3, pp.90-96, 1999.

[3] S.H.Yeo, “A Multi-Pass Optimization Strategy for CNC Lathe Operations”, International Journal of Production Economics, Vol.40, pp.209-218, 1995.

[4] W. S. Lin, B. Y. Lee, and C. L. Wu, “Modeling the Surface Roughness and Cutting Force for Turning”, Journal of Materials Processing Technology, Vol.108, pp286-293, 2001.

[5] D. Gillibrand, M. Sarwar, and C. T. Pierce “The Economic Benefit of Finish Turning with Coated Carbide”, Surface and Coatings Technology, Vol.86-87, pp.809-813, 1996.