Int J Adv Manuf Technol (2006) 31: 164–174 DOI 10.1007/s00170-005-0180-0 ORIGINAL ARTICLE Ting-Cheng Chang . Feng-Che Tsai . Jiuan-Hung Ke Data mining and Taguchi method combination applied to the selection of discharge factors and the best interactive factor combination under multiple quality properties Received: 18 March 2004 / Accepted: 27 May 2005 / Published online: 16 March 2006 # Springer-Verlag London Limited 2006 Abstract This research uses data mining to effectively analyze and confirm that the factors affecting discharge for electrical discharge machining (EDM) are, pulse-on time, pulse-off time, open discharge voltage, or interval voltage. The Taguchi method is used to experiment with electrodes of the same size and different shapes, based on key factors acquired under the consideration of the interaction between factors, and analyze the experimental results respectively with average values and S/N ratio. Eventually, the research will determine and calculate the best combination of factors of each quality property (machining and reaming amount, surface roughness, electrode corner loss, and material re- moval rate (MRR)) by analyzing predicated values and prove it with experimental data. The experiment proves that the interaction between factors really exists and that the best combination of factors acquired has favorable effects and credibility. Keywords Data mining . Discernibility matrix . Electrical discharge machining . K-means . Rough set . Taguchi method 1 Introduction Electrical discharge machining, developed by Mr. And Mrs. D.R. Larzarlinko from Russia in 1943, is a machining method used to explode and melt remnants electrically. In the beginning, electrical discharge machining is developed to get rid of broken taps, drills, or other special tools. Due to its particular machining method, electrical discharge machining is widely used in mold manufacturing, wire electrical discharge machining, and other precision industries. Unlike traditional cutting methods, electrical discharge machining is not limited by the hardness of workpieces. The quality of electrical discharge machining however, is not ensured because discharge control factors constitute by different planning. In addition, the machining speed and control factor setting cannot be controlled effectively. These are the biggest challenges in electrical discharge machining. Manufacturers try to ascertain discharge control factors to improve the machining quality based on their accumu- lated operation experiences, operation manuals, or failed attempts. Academic researchers have probed into how to achieve a process of optimizing multiple quality properties as conducting discharge machining using various methods in recent years. Here are some researches to be referred to: Osyczka et al. [1] address multi-criterion optimization to improve the quality properties of metal removal rate, sur- face roughness, and electrode wastage. Poluyanov [2] methods to probe into the effects of electrical discharge machining caused by electrodes’ surface areas and shapes, by analyzing in the view of the chip-relief routes and represents that electrodes with the same surface areas and different shapes produce different maximum current cir- culation capacity. Lin et al. [3, 4] analyze the best factor combination in combination of Taguchi method and fuzzy logic, to improve the quality properties of MRR and electrode wastage, and to establish an algorithm for mul- tiple quality properties to achieve obscure design targets. Lin et al. [5–7] develop a set of algorithms to improve the quality properties of MRR, surface roughness, and elec- trode wastage with a combination of Taguchi method and gray relational analysis. Although these researches com- bine with various methods to improve discharge quality, each paper lacks the basis to influence discharge quality factors. The quality property selection is limited to MRR, surface roughness, and electrode wastage, ignorant of the amount of machining and reaming. Since electrodes lead workpieces to produce different discharge sizes after the T.-C. Chang (*) Department of Management Information, Ling Tung University of Technology, #1 Ling Tung Road, Taichung, 408 Taiwan e-mail: weny@ms6.hinet.net Tel.: +886-4-36018082 Fax: +886-4-23821912 F.-C. Tsai . J.-H. Ke Department of Mechanical Engineering, National Central University, Taichung, Taiwan