Obrabotka Metallov. 2016 no. 2(71)

ОБРАБОТКА МЕТАЛЛОВ № 2 (71) 2016 38 ОБОРУДОВАНИЕ. инструменты OBRABOTKAMETALLOV (METAL WORKING AND MATERIAL SCIENCE) N 2 (71), April – June 2016, Pages 28–40 An assessment of cutting abilities of boron nitride high porous wheels while pendulum grinding of elements made of titanium alloy VT20 using the artificial neural network Soler Ya. I. , Ph.D. (Engineering), Associate Professor, e-mail: solera@istu.irk.ru Mai D. S. , Ph.D. student, e-mail: mdsmm07@gmail.com Nguyen V.L. , Ph.D. student, e-mail: nhatle007@gmail.com National Research Irkutsk State Technical University, 83 Lermontova st., Irkutsk, 664074, Russian Federation Abstract The high porous wheels (HPW) made of cubic boron nitride (CBN) are used to improve the grinding efficiency of titanium alloys. The high dimensional pores in these tools allow to reduce the blunting on the HPW working surfaces and to avoid appearance of the grinding burns and cracks on the ground surfaces. In this paper the cutting ability (CA) of six HPW from CBN was carried out while grinding parts from VT20 alloy. The studied HPW have the constant grain B126, but their other characteristics were varied: by the CBN marks - from CBN30 to LCV 50, by the wheel hardness - from L (medium soft) to O (medium hard) and by the pore-forming agent - from KF25 to KF40. The CA of the tools are measured by the high-rise indicator (R a , R max ) and the stepping indicator of roughness (S m ) (GOST 25142-82). The analysis of the observations is leaded using statistical approaches, because the grinding process has a stochastic character. The nonparametric statistical method is used on the basis of the experimental data results of testing on the homoscedasticity and the normality of distributions. In this case, the measure of position is the medians and the measures of dispersion are the quartile latitude (QL). For the complex assessment of the HPW’s CA the simulation in the artificial neural network in «STATISTICA Neural Networks» package was carried out. By its results, it was established that the HPW LCV50 B126 100 MV K27-KF40 with a rating “very good” provides the highest surface quality in grinding flat parts of the VT20. Moreover it was found that the VT20 is ground better by the HPW made from the grain B126 with high grain strength 50, low hardness (L, M) and the most pore-forming (KF40). The «STATISTICA Neural Networks» package also has an option to predict the network sensitivity to input variables. In this case, the order of decreasing the influence on the quality assessment of the part surface are R a , R max , QL ( R max ), QL ( S m ), QL ( R a ) and S m . Keywords grinding, titanium alloy VT20, statistics, artificial neural network, sensitivity. DOI: 10.17212/1994-6309-2016-2-28-40 References 1. Nosenko V.A., Nosenko S.V. Tekhnologiya shlifovaniya metallov [The grinding technology of metals]. Staryi Oskol, TNT Publ., 2013. 616 p. ISBN 978-5-94178-373-1 2. Sayutin G.I., Tatarinov I.P. Vybor materiala kruga pri shlifovanii titanovykh splavov [The choice of material for grinding wheel titanium alloys]. Stanki i instrument – Soviet Engineering Research , 1985, no. 7, pp. 21–22. (In Russian) 3. Kremenj Z.I., Zubarev Yu.M., LebedevA.I. Vysokoporistye krugi iz el’bora i ikh primenenie pri shlifovanii vysokoplastichnykh splavov [High-porous CBN vitrified wheels and their application in grinding of high-ductile al- loys]. Metalloobrabotka – Metalworking, 2009, no. 3 (51), pp.  2–5. 4. Il’in A.A., Kolachev B.A., Pol’kin I.S. Titanovye splavy: sostav, struktura, svoistva : spravochnik [Titanium alloys: composition, structure, properties. Reference book]. Moscow, VILS-MATI Publ., 2009. 520 p. 5. Nguyen D.M. Kompleksnoe issledovanie zadachi klassifikatsii s primeneniem nechetkikh modelei i raspredel- ennykh vychislenii. Diss. kand. tekhn. nauk [A comprehensive study of the classification problem using fuzzy models and distributed calculations. PhD eng. sci. diss.]. Irkutsk, 2014. 142 p.